are dba/2 mice associated with schizophrenia-like endophenotypes? a behavioural contrast with...

ORIGINAL INVESTIGATION

Are DBA/2 mice associated with schizophrenia-likeendophenotypes? A behavioural contrast with C57BL/6 mice

Philipp Singer & Joram Feldon & Benjamin K. Yee

Received: 18 November 2008 /Accepted: 7 May 2009 /Published online: 30 May 2009# Springer-Verlag 2009

AbstractRationale Due to its intrinsic deficiency in prepulseinhibition (PPI), the inbred DBA/2 mouse strain has beenconsidered as an animal model for evaluating antipsychoticdrugs. However, the PPI impairment observed in DBA/2mice relative to the common C57BL/6 strain is confoundedby a concomitant reduction in baseline startle reactivity. Inthis study, we examined the robustness of the PPI deficitwhen this confound is fully taken into account.Materials and methods Male DBA/2 and C57BL/6 micewere compared in a PPI experiment using multiple pulsestimulus intensities, allowing the possible matching ofstartle reactivity prior to examination of PPI. The knownPPI-enhancing effect of the antipsychotic, clozapine, wasthen evaluated in half of the animals, whilst the other halfwas subjected to two additional schizophrenia-relevantbehavioural tests: latent inhibition (LI) and locomotorreaction to the psychostimulants—amphetamine andphencyclidine.Results PPI deficiency in DBA/2 relative to C57BL/6 micewas essentially independent of the strain difference inbaseline startle reactivity. Yet, there was no evidence thatDBA/2 mice were superior in detecting the PPI-facilitatingeffect of clozapine when startle difference was balanced.Compared with C57BL/6 mice, DBA/2 mice also showedimpaired LI and a different temporal profile in theirresponses to amphetamine and phencyclidine.Conclusion Relative to the C57BL/6 strain, DBA/2 micedisplayed multiple behavioural traits relevant to schizophrenia

psycho- and physiopathology, indicative of both dopaminer-gic and glutamatergic/N-methyl-D-aspartic acid receptordysfunctions. Further examination of their underlying neu-robiological differences is therefore warranted in order toenhance the power of this specific inter-strain comparison asa model of schizophrenia.

Keywords Amphetamine . Behavioural genetic .

Latent inhibition .Mouse . Phencyclidine .

Prepulse inhibition . Schizophrenia

Introduction

Behavioural phenotypic divergence between strains oflaboratory rodents is relevant to behavioural genetics andprovides an opportunity to develop animal models ofspecific psychopathological traits. One approach is todevelop selective breeding (e.g. Logue et al. 1997;McCaughran et al. 1997; Schwabe et al. 2007), whilstanother is to compare existing, usually inbred strains withrelatively stable genetic backgrounds (Bullock et al. 1997;Crawley et al. 1997; Paylor and Crawley 1997). The inbredDBA/2 mouse strain has attracted interests recently in itspotential relevance to psychosis-related traits in comparisonto another common inbred mouse strain—the C57BL/6strain—as the reference line (Olivier et al. 2001; Spielewoyand Markou 2004). DBA is the oldest recorded inbredmouse strain originating from Little in 1909, and substrainsDBA/1 and DBA/2 were established in 1929–1930, both ofwhich are now maintained in the Jackson Laboratory (seeMouse Genome Informatics at http://www.informatics.jax.org/external/festing/mouse/docs/DBA.shtml). Since the1960s, strain comparison studies have reported variousnotable behavioural phenotypes in the DBA/2 substrain,

Psychopharmacology (2009) 206:677–698DOI 10.1007/s00213-009-1568-6

P. Singer : J. Feldon :B. K. Yee (*)Laboratory of Behavioural Neurobiology,Federal Institute of Technology Zurich,Schorenstrasse 16,8603 Schwerzenbach, Switzerlande-mail: [email protected]

http://www.informatics.jax.org/external/festing/mouse/docs/DBA.shtml

http://www.informatics.jax.org/external/festing/mouse/docs/DBA.shtml

as catalogued in the Mouse Genome Informatics database(see link above). A number of cognitive traits revealed inDBA/2 mice, such as spatial memory deficits, have led tothe suggestion that they may represent a model ofhippocampal dysfunction (Ammassari-Teule et al. 1995;Paylor et al. 1993).

Of particular relevance to psychotic traits is theconsistent reporting of weak prepulse inhibition (PPI)expression and related physiological gating in DBA/2 mice(e.g. Bortolato et al. 2007; Bullock et al. 1997; Logue et al.1997; McCaughran et al. 1997; Olivier et al. 2001; Stevenset al. 1996; Stevens and Wear 1997). PPI represents a formof sensorimotor gating that is deficient in schizophreniapatients, and it can be readily demonstrated in humans aswell as in animals using the ubiquitous acoustic startlereflex (Hoffman and Searle 1965). PPI is commonlydemonstrated by the reduction in the startle reaction to aloud pulse stimulus when it is shortly preceded by a brief,non-startling prepulse stimulus. It is therefore a powerfultranslational paradigm for dysfunction in pre-attentiveinformation processing and sensorimotor gating whichexists in various psychiatric disorders, including schizo-phrenia (Braff et al. 2001; Csomor et al. 2009; Geyer et al.2002; Swerdlow and Geyer 1998; Swerdlow et al. 2008). Anumber of comparative studies between inbred mousestrains have revealed considerable divergence in PPIexpression (e.g. Bullock et al. 1997; Geyer et al. 2002;Logue et al. 1997; McCaughran et al. 1997; Paylor andCrawley 1997) and led to the suggestion that mouse strainswith low basal levels of PPI may be useful in studying theneural basis of PPI deficiency without the requirement ofadditional pharmacological interventions to induce a PPIdeficit (see McCaughran et al. 1997). Olivier et al. (2001)further proposed more specifically that the weak PPIphenotype observed in the DBA/2 substrain (relative toother mouse strains) may serve as a model to screen forantipsychotic drugs—a recommendation that has beenadopted in a number of recent psychopharmacologicalstudies of novel antipsychotics, including histamine H3receptor antagonists (Browman et al. 2004; Fox et al. 2005)and glycine transporter 1 inhibitors (Boulay et al. 2008;Depoortère et al. 2005; Kinney et al. 2003).

However, past studies have largely overlooked theconcomitant change in startle reaction magnitude. InDBA/2 mice, the reported PPI deficits are confounded bya pronounced reduction in startle reaction (e.g. Bortolato etal. 2007; Crawley et al. 1997; McCaughran et al. 1997;Paylor and Crawley 1997). This is of significant concernbecause differences in the baseline startle reactivity canseriously complicate or confound the interpretation of theresults obtained on PPI (Csomor et al. 2006, 2008;Hoffman and Searle 1965; Swerdlow et al. 2000; Yee etal. 2005). The dependency of PPI magnitude on baseline

startle reactivity has been demonstrated in rodents andhumans (Csomor et al. 2006, 2008; Hince and Martin-Iverson 2005; Yee et al. 2005)—a finding that has beenincorporated also by computational models of PPI (Sandnerand Canal 2007; Schmajuk and Larrauri 2005). The presentstudy was undertaken to initially address the confoundingphenotype on startle reactivity in comparison with C57BL/6mice using a more stringent parametric design of PPIassessment incorporating multiple pulse intensities (Csomoret al. 2006, 2008; Yee et al. 2005). With this PPI design, wewent on to investigate if the DBA/2 substrain may enjoysuperior sensitivity (compared with C57BL/6 mice) in thedetection of atypical neuroleptic drugs using clozapine as thereference compound.

The present study also includes comparisons betweenDBA/2 and C57BL/6 mice in another behavioural paradigmof psychosis-related attentional deficits, namely latentinhibition (LI). Unlike PPI, LI refers to a form of learnedattentional control by which repeated exposures to aninconsequential stimulus lead to a loss of associability ofthe stimulus, as shown by its weakened ability to enter intoassociation with other significant stimuli in Pavlovian andinstrumental conditioning paradigms (Lubow and Moore1959). LI is also deficient in some subsets of schizophreniapatients (e.g. Baruch et al. 1988; Gray et al. 1992; Williamset al. 1998), and it has been suggested that a distinctionbetween pathological traits associated with positive andnegative symptoms of schizophrenia can be achieved byappropriate parametric variation of the LI paradigm (seeWeiner 2003). Such manipulation was included here, thusallowing an important extension to previous studies of LIexpression in DBA/2 mice (Baarendse et al. 2008; Gouldand Wehner 1999; Restivo et al. 2002). Finally, we alsoevaluated in the comparison between DBA/2 and C57BL/6mice whether the former would exhibit pharmacologicallyinduced phenotypes that are in keeping with the dopamineand glutamate/N-methyl-D-aspartic acid (NMDA) hypothe-ses of schizophrenia (c.f. Coyle 2006; Carlsson 1988; Javitt2007; Snyder 1976). To this end, the animals weresubjected to acute challenge either of an indirect dopaminereceptor agonist, amphetamine, or the non-competitiveNMDA receptor antagonist, phencyclidine (PCP).

Materials and methods

Subjects

Adult male C57BL/6 (n=36) and DBA/2 (n=40) mice wereobtained from the Charles River Laboratories (Germany).They were kept in groups of four to five in Makrolon® typeIII cages (Techniplast, Milan, Italy) and housed in ananimal vivarium under a reversed 12/12-h light–dark cycle

678 Psychopharmacology (2009) 206:677–698

(lights on 2000–0800 hours) with ad lib food and waterthroughout the entire experimental period. The mice wereacclimatised to the laboratory housing conditions for2 weeks before the commencement of behavioural testingat an age of approximately 12 weeks. The tests were allconducted in the dark phase of the light–dark cycle. Allprocedures performed on the animals had been approved bythe Zurich Cantonal Veterinary Office, in accordance withthe Animal Protection Act of Switzerland (1978), theEuropean Council Directives 86/609/EEC on animal exper-imentation and the Principles of Laboratory Animal Care(NIH publication no. 86-23, revised 1985).

Overall experimental design

First of all, in experiment 1, PPI was evaluated in all theanimals, after which they were subdivided into two balancedcohorts in terms of mean startle reactivity and PPI expres-sion. One cohort then was subjected to another PPIexperiment in which the efficacy of clozapine to modifyPPI expression was examined with a 2×2 (strain × drug)factorial design (experiment 2; n=9 or 10 per cell). From thesecond cohort, 16 DBA/2 mice and 16 C57BL/6 mice wererandomly selected and further subdivided for two separateLI experiments, each using a different number of stimuluspre-exposures (experiments 3A and 3B; n=4 per cell). Dueto time limitations (for the experiments to be completed in2 days) and the availability of just four shuttle boxes,experiments 3A and 3B together were carried out with only32 out of the available 38 animals. Following the completionof experiment 3, the animals were evaluated in locomotorreaction to amphetamine or PCP in comparison with salinevehicle treatment (experiment 4; n=5 or 6 per cell). Theoverall experimental plan including sequence, duration ofeach test, number of rest days between tests and the numberof subjects are summarised in Fig. 1.

Prepulse inhibition of the acoustic startle reflex

Apparatus The apparatus consisted of four acoustic startlechambers for mice (SR-LAB, San Diego Instruments, SanDiego, CA, USA) each containing a non-restrictive clearPlexiglas cylindrical enclosure attached to a horizontalmobile platform, which, in turn, rested on a solid baseinside a sound-attenuated isolation cubicle. A high-frequency loudspeaker, mounted directly above the animalenclosure inside each cubicle, produced a continuousbackground noise of 65 dBA, together with the variousacoustic stimuli in the form of white noise. The forcesacting on the Plexiglas enclosure caused by the whole-bodystartle response of the animal were converted into analoguesignals by a piezoelectric unit attached to the platform.These signals were digitised and stored by a computer. A

total of 130 readings were taken at 0.5-ms intervals (i.e.spanning 65 ms), starting at the onset of the startle stimulusin pulse-alone and prepulse-plus-pulse trials and at theonset of the prepulse stimulus in prepulse-alone trials. Theaverage amplitude over the 65 ms was used to determinethe stimulus reactivity. The sensitivity of the stabilimeterwas routinely calibrated to ensure consistency betweenchambers and across sessions.

Procedures Here, we adopted the procedures and testingparameters developed by Yee et al. (2005). PPI wasassessed in a test session lasting for approximately 40 minin which the subjects were presented with a series ofdiscrete trials comprising a mixture of four types of trial.These included pulse-alone trials, prepulse-plus-pulse trials,prepulse-alone trials and trials in which no discretestimulus, other than the constant background noise, waspresented (“no-stimulus” trials). A reduction of startlemagnitude in prepulse-plus-pulse trials relative to that inpulse-alone trials (of the same pulse intensity) constitutesPPI. Pulse stimuli of 40-ms duration and of three differentintensities were employed: 100, 110 and 120 dBA. Prepulsestimuli of 20-ms duration were employed with threedifferent intensities: 71, 77 and 83 dBA (corresponding to6, 12 and 18 decibel units above background, respectively).In the prepulse-plus-pulse trials, a stimulus onset asynchro-ny of 100 ms was used.

A session began when the animals were placed into thePlexiglas enclosure. They were acclimatised to the appara-tus for 2 min and then presented with three blocks of trials.The first block comprised six pulse-alone trials: two trialsof each of the three possible pulse intensities. These trialsserved to habituate and stabilise the animals' startleresponse and were analysed separately. In the intermediateblock, the animals were presented with ten series of discretetest trials. Each series consisted of the following trials: threepulse-alone trials (100, 110 or 120 dBA), three prepulse-alone trials (+6, +12 or +18 decibel units above background)and nine possible combinations of prepulse-plus-pulse trials(three levels of prepulse × three levels of pulse) and one no-stimulus (or background alone) trial. The 16 discrete trialswithin each series were presented in a pseudorandom order,with a variable inter-trial interval of a mean of 15 s (rangingfrom 10 to 20 s). The session was concluded with a finalblock of six consecutive pulse-alone trials as in the firstblock.

Drug administration In experiment 2 (see Fig. 1), theanimals were administered with either clozapine or vehiclesolution 45 min prior to PPI testing. Allocation of subjectsto drug conditions was counterbalanced with respect tobaseline PPI and startle reactivity scores obtained inexperiment 1. Thus, the repeated use of animals from

Psychopharmacology (2009) 206:677–698 679

experiments 1 to 2 offered a distinct advantage. Thereplication of the baseline difference in PPI expressionbetween strains in experiment 2 (see later) confirmed thatthe results obtained in experiment 2 were unlikely to bebiased by the repeated use of the same animals as such.Clozapine (obtained from Sigma-Aldrich, Germany) wasdissolved in 0.1 N HCl in physiological saline solution andthen neutralised to pH5.5 with Na2CO3 to achieve thedesired dose of 1 mg/kg. For the vehicle treatment,physiological saline/lactic acid (pH5.5) was prepared. Allsubstances were freshly prepared on the day of injectionand administered in a volume of 5 ml/kg body weight viathe intraperitoneal route (i.p.).

Latent inhibition in two-way active avoidance learning

Apparatus The apparatus comprised four identical two-wayshuttle boxes (model H10-11M-SC; Coulbourn Instru-ments) each housed within a sound-attenuating chest, asdescribed fully elsewhere (Chang et al. 2007; Yee et al.2006). With internal dimensions of 35.5×17×21.5 cm, eachbox was divided into two identical compartments by analuminium wall with an interconnecting opening (6.77×7.7 cm), allowing free movement of the animal from onecompartment to the other (i.e. a shuttle response). The gridfloor was made of stainless steel rods (diameter, 0.4 cm;spaced at 0.7 cm centre-to-centre) through which electricshocks (0.3 mA) generated by a shock module (model H10-1M-XX-SF; Coulbourn Instruments) could be delivered.The conditioned stimulus (CS) was an 85 dBA white noisegenerated by a tone module (model E12-02; Coulbourn

Instruments) mounted behind the shuttle box. Shuttleresponse was detected by a series of photocells (H20-95X; Coulbourn Instruments) mounted on the sides of bothcompartments. The boxes were illuminated during the entireexperimental session by two diffused light sources (1.1 W),one mounted on the end wall of each compartment 19 cmabove the grid floor. Background noise was provided by aventilation fan mounted at the back of each shuttle box.

Procedures Weiner (2003) emphasises that abnormality inLI may assume the form of an attenuation (against thepresence of strong LI in controls) or abnormal persistence(against the absence of LI in controls). It would benecessary to examine both these possibilities. Here, wemanipulated the magnitude of LI in the C57BL/6 group byvarying the number of stimulus pre-exposures. With 100stimulus pre-exposures, robust LI was obtained in C57BL/6mice, whilst 50 pre-exposures yielded no significant LI inthis strain of mice in two-way active avoidance (P. Singer,W-N. Zhang, J. Feldon and B.K. Yee, unpublished data).Otherwise, the procedures for the two experiments wereidentical and followed a 2×2 factorial design with thebetween-subjects factors: strain (DBA/2 vs. C57BL/6) andpre-exposure (PE vs. nPE). These factors were counter-balanced across the four shuttle boxes. Each experimentcomprised two phases, pre-exposure and conditioning,conducted 24 h apart.

On the pre-exposure day, the animals in the pre-exposed(PE) condition received either 50 or 100 exposures to theto-be-conditioned white noise (experiments 3A and 3B,respectively) presented at a variable inter-stimulus intervalof 40 s (ranging from 25 to 55 s), whilst the non-pre-

Fig. 1 Summary of the overall experimental design in the present study. PE pre-exposure condition, nPE non-pre-exposure condition, amphamphetamine, PCP phencyclidine. Days of rest between experiments indicated just above or below the arrows


exposed (nPE) subjects spent an equivalent amount of timein the shuttle boxes without any programmed events exceptfor the house light.

On the next day, the animals were returned to the sameboxes and received 100 active avoidance trials presentedaccording to a random interval schedule (mean 40 s,ranging from 25 to 55 s). A trial began with the onset ofthe white noise CS. If the animal shuttled within 5 s of CSonset, the CS was terminated and the animal did not receiveany foot shock: This was scored as an avoidance response.Avoidance failure led to the immediate delivery of a footshock (presented in conjunction with the white noise CS)that could last for a maximum of 2 s. A shuttle responseduring this period would terminate both the CS and the footshock, and an escape response would be recorded.Otherwise, when an animal failed to shuttle in the presenceof the shock, an escape failure was recorded. Shuttleresponses performed during the inter-trial intervals wereall without programmed consequence, but these wererecorded and analysed separately. Avoidance learning wasindexed by the number of successful avoidance responsesacross successive blocks of ten trials. Retarded learning inthe PE subjects relative to the respective nPE controlswould constitute LI.

Locomotor response to psychostimulant drugs in the openfield

Apparatus The apparatus consisted of eight identical openfield arenas constructed from plastic laminated wood with awaterproof white finish. Each arena measured 40×40 cm insurface area and was surrounded on all sides by a 30-cm-high wall. The experiment was conducted in two separatetesting rooms with diffused dim lighting (30 lx). Each roomhoused four open field arenas positioned directly under adigital camera that captured images from all four arenas at arate of 5 Hz. The images were transmitted to a personalcomputer (PC) running the Ethovision (Noldus Technology,The Netherlands) tracking system, and locomotor activitywas indexed by the cumulative displacement of the centreof gravity of the subject's surface area over successiveframes, expressed in 10-min bins.

Procedures The allocation of animals between the twotesting rooms and amongst the eight arenas was counter-balanced with respect to strains and drug conditions.Immediately following the systemic injection (amphet-amine, PCP or saline vehicle), each animal was gentlyplaced in the centre of the appropriate arena and allowed toexplore undisturbed for 2 h. They were not previouslyhabituated to the arena because we intended to capture theunique profile of psychostimulant drug in association with

environmental novelty—an aspect that is critical to bothlatent inhibition and prepulse inhibition on theoreticalgrounds (e.g. Schmajuk and Larrauri 2005; Schmajuk etal. 1996). Afterwards, they were returned to the home cageand the arenas cleansed with water and dried prior to thenext squad. All solutions for injection were freshly preparedon the day of testing and administered in a volume of5 ml/kg body weight via the i.p. route. Amphetamine(obtained from Sigma-Aldrich, Switzerland) and phencycli-dine (PCP, donated by the National Institute for DrugAddiction, USA) were dissolved in sterile physiological salineto achieve the required doses of 2.5 and 5 mg/kg, respectively.

Statistical analysis

All data were analysed by parametric analysis of variance(ANOVA) using the between-subject factors, strain, incombination with drug and pre-exposures as appropriate.Additional within-subject factors (e.g. 10-min bins, ten-trialblocks) were also included according to the nature of theconsidered dependent variables. To assist interpretation ofthe statistical outcomes, significant effects were furtherinvestigated by supplementary restricted analyses applied toa subset of the data or by pairwise comparisons based onthe associated error terms taken from the overall ANOVAwhenever appropriate. To better conform to the normalityand variance homogeneity assumptions of parametricANOVA, logarithmic transformation (indicated as “ln-transformed” in the text and figures) was applied to thereactivity data in the prepulse inhibition experiment [ln(reactivity score + e) − 1] (see Csomor et al. 2008). Asquare-root transformation was likewise applied to thenumber of inter-trial interval (ITI) crossings in the activeavoidance experiment as well as to the activity measure inthe open field experiment. All statistical analyses werecarried out using SPSS for Windows (release 13, SPSS,Chicago IL, USA) implemented on a PC running theWindows XP (SP3) operating system.

Results

Experiment 1: comparison of baseline PPI expressionin DBA/J2 and C57BL/6 mice

Here, both the expression of PPI and the acoustic startlereflex were comprehensively evaluated using the parametricdesign adopted from Yee et al. (2005) because of theanticipated between-strain difference in startle reactivity.Different data sets were derived from the test session andsubjected to separate analyses to address specific processesas follows: (1) Startle reactions obtained in the first and last


blocks of trials consisting only of pulse-alone trials wereused to assess startle habituation. (2) Evaluation of thereaction to pulse-alone trials presented in the intermediateblock of trials, when they were intermixed with differenttrial-types, allowed the assessment of startle reactivity thatwould be expected to affect PPI assessment. (3) PPIinhibition or reduction of startle reaction to the pulsestimulus induced by a preceding prepulse stimulus wasassessed in an analysis of all pulse-alone and prepulse-plus-pulse trials presented in the intermediate trial block. (4) PPIexpression as percent reduction of the startle response (%PPI) was included here for the purpose of comparison withprevious studies that rely solely on this indexation of PPI,although there are serious reservations for its applicationwhen startle reactivity differed between treatment groups(Csomor et al. 2008; Swerdlow et al. 2000; Yee et al. 2005).(5) Whenever appropriate and feasible, additional compar-isons of PPI between strains were conducted with anattempt to balance the reactivity on pulse-alone trials bycontrasting the data derived from one pulse intensity in onestrain to the data obtained with another pulse intensity inthe other strain. This entire analytic approach was adoptedin experiments 1 and 2. All reactivity scores were subjectedto a logarithmic transformation (ln-transformed) prior tostatistical analysis as explained in “Materials and methods”(also see Csomor et al. 2008; Yee et al. 2005).

Startle habituation

The first and the last six trials of the test session comprisedonly pulse-alone trials (two trials per intensity: 100, 110 or120 dBA). Comparison between these two blocks of sixtrials allowed the assessment of long-term startle habitua-tion. As illustrated in Fig. 2a, a marked decrease in thestartle reaction from the first to the last blocks was apparentin the DBA/2 mice, but not in the C57BL/6 mice. Whereasstartle habituation was present in DBA/2 mice regardless ofpulse stimulus intensity, C57BL/6 mice displayed noobvious tendency of habituation. If anything, C57BL/6mice showed a tendency of sensitisation to the pulsestimulus at the two lower intensities from the first to thelast blocks. These impressions were confirmed by a 2×2×3(strain × blocks × pulse intensity) split-plot ANOVA, whichyielded a significant strain by blocks interaction [F(1,47)=51.64, p<0.001], and the three-way interaction [F(2,148)=9.44, p<0.001]. These were accompanied by a main effectof blocks [F(1,74)=24.56, p<0.001], of pulse intensity[F(1,74)=95.28, p<0.001] and of strain [F(1,74)=8.57, p=0.005], indicating the overall presence of a habituationeffect and the monotonic dependency of startle reaction onpulse intensity and that the two strains differed in startlereactivity. Supplementary analyses restricted to each strainconfirmed the presence of a significant habituation effect in

the DBA/2 strain [blocks: F(1,39)=66.38, p<0.001], butnot in the C57BL/6 strain [F(1,35)=2.92, p=0.10].

Startle reaction in pulse-alone trials in the main block

The reaction obtained in pulse-alone trials in the main block(i.e. as apart from the first and last blocks of pulse-alonetrials described above) was used in the assessment of PPIand was therefore analysed separately. The results confirmedthe overall difference between the two mouse strains and thatthis strain difference was proportional to the intensity of thepulse stimulus (Fig. 2b, shaded region). These interpretationswere supported by a 2×3 (strain × pulse intensity) split-plotANOVA, which yielded a significant main effect of strain[F(1,74)=19.72, p<0.001], pulse intensity [F(2,148)=169.75, p<0.001] as well as their interaction [F(2,148)=41.23, p<0.001]. Pairwise comparisons based on the errorvariance associated with the significant interaction termconfirmed the presence of a significant strain differenceunder the 110- and 120-dB conditions (p<0.0001), but not inthe lower intensity (100 dB) condition.

Prepulse inhibition: pulse-alone vs. prepulse-plus-pulsetrials

PPI is defined as the reduction of startle reaction to thepulse stimulus when it is preceded by a prepulse stimulus,relative to when it is not. Here, this was statisticallyassessed using a within-subject factor that incorporatedboth trial types as well as prepulse intensity, +0, +6, +12and +18 dB, with +0 effectively referring to the pulse-alonecondition. This is the recommended method in analysingtreatment effects known to be confounded by a differencein startle reactivity (Swerdlow et al. 2000).

As illustrated in Fig. 2b, the presence of PPI was evidentin both DBA/2 and C57BL/6 mice and across the three pulseintensities. Increasing prepulse intensity (from “+0” repre-senting pulse-alone condition) led to a monotonic reductionin startle magnitude, and this reduction appeared morepronounced in conditions with higher pulse intensity. Thesteepness of the individual curves depicted may be inter-preted as the efficacy of the increasingly intense prepulsestimulus to inhibit more strongly the startle response elicitedby the pulse stimulus (see Yee et al. 2004b). Thus, thedepiction represented by Fig. 2b suggests that DBA/2 micewere in general less responsive to the prepulse.

A 2×3×4 (strain × pulse intensity × prepulse intensity)split-plot ANOVA revealed a significant main effect ofpulse intensity [F(2,148)=229.05, p<0.001], as expected.The presence of a significant prepulse intensity effect[F(3,222)=240.80, p<0.001] and its interaction with pulseintensity [F(6,444)=10.26, p<0.001] confirmed the pres-ence of PPI and its dependency on pulse intensity. The


analysis also revealed significant differences betweenmouse strains. The significant main effect of strain [F(1,74)=11.65, p<0.05] and its interaction with pulseintensity [F(2,148)=34.28, p<0.001] are in keeping withthe impressions obtained from the preceding analysis:DBA/2 mice showed reduced overall startle reactivity,and this strain difference was dependent on pulseintensity. Concerning the comparison of PPI expressionbetween strains, the analysis also revealed a significantstrain by prepulse intensity interaction [F(3,222)=24.05,p<0.001] and the three-way interaction [F(6,444)=10.26,

p=0.001]. These effects support the conclusion that inDBA/2 mice, responsiveness to the prepulse was attenu-ated, and this attenuation was dependent on pulseintensity, apparently more pronounced at higher thanlower pulse intensities.

Startle-matched and cross-pulse intensity analysis of PPI:pulse-alone vs. prepulse-plus-pulse trials

Because response in pulse-alone trials was highly compa-rable between strains at the lowest pulse (100 dB), a

Fig. 2 Summary of the results in experiment 1. Based on the PPI testsession, the following data sets were derived and separately analysed.a Comparison between the first and last blocks of the test sessioncomprising pulse-alone trials only with two trials of each pulseintensity (100, 110 and 120 dBA, represented by increasing shades ofgrey). Results of the two strains were plotted separately (left, DBA/2;right, C57BL/6). b Average reactivity scores (following logarithmictransformation) across 12 different types of trial definition obtained inthe intermediate block of the test session: pulse-alone trials (emphas-ised by the shaded rectangular background) are denoted as “+0 dB”in the abscissa. The intensity of the prepulse stimulus in prepulse-

plus-pulse trials was also denoted along the abscissa as +6, +12 and+18 decibel units above the constant background noise level of65 dBA. Three separate plots were illustrated corresponding to the threepulse stimulus intensities: 100, 110 and 120 dBA, as indicated at the top.c Response (again following a logarithmic transformation) obtained inprepulse-alone trials (of three intensities) and in comparison to “no-stimulus” trials. d Results in percent prepulse inhibition (%PPI) derivedfrom the same data set as illustrated in b. Circles, C57BL/6 strain;squares, DBA/2 strain (regardless of fillings). All values are means withassociated standard errors (±SE) based on the respective ANOVAconducted using SPSS for Windows (version 13)


restricted ANOVAwas performed on data derived from thisparticular test condition. As expected, the main effect ofstrain was no longer significant (F < 1). The PPI effect wasconfirmed by the presence of a significant main prepulseeffect [F(3,222)=45.11, p<0.001]. However, evidence of aPPI deficit in the DBA/2 mice was weaker: The strain byprepulse intensity interaction failed to achieve statisticalsignificance [F(3,222)=1.90, p=0.13]. Only the quadratictrend of this interaction remained statistically significant[F(1,74)=5.00, p<0.05]. Hence, there is tentative evidencethat DBA/2 mice exhibited a PPI deficiency relative toC57BL/6 mice independent of the confounding reduction instartle reactivity.

As a further and more stringent test of this assertion,we next selected for comparison between the two strainsthe highest pulse intensity (120 dB) data set from DBA/2 mice and the lowest pulse intensity (100 dB) data setfrom C57BL/6 mice. This effectively reversed thedifference in pulse-alone reaction between strains. A2×4 (strain × pre-pulse intensity) split-plot ANOVA ofthis data set revealed again a significant interactionbetween strain and prepulse [F(3,222)=4.40, p<0.005],which still supports the interpretation that the startleresponse in DBA/2 mice was less sensitive to theinhibition by the prepulse. The significant main effect ofstrain [F(1,74)=38.06, p<0.001] in this analysis nowrefers to overall startle response being higher in the DBA/2 mice. The main effect of prepulse was also significant[F(3,222)=42.64, p<0.001], indicating the overall pres-ence of PPI.

Prepulse inhibition: percent startle inhibition

In conformity to the common method of PPI indexationby percent startle reduction calculated with respect to thecorresponding pulse-alone condition, we also presentedour data using the expression %PPI = (pulse-alone −prepulse-plus-pulse)/pulse-alone × 100%) using the un-transformed reactivity scores for each of the nine (threeprepulse intensities × three pulse intensities) prepulse-plus-pulse conditions (see Fig. 2d). A 2×3×3 (strain ×pulse intensity × prepulse intensity) split-plot ANOVAof %PPI revealed an overall strain effect [F(1,74)=38.40,p<0.001], which is suggestive of a PPI deficit in theDBA/2 mice. Dependency of this deficit on pulseintensity was supported by the strain by pulse intensityinteraction [F(2,148)=5.79, p<0.005], which was accom-panied by a significant main effect of pulse intensity[F(2,148)=8.00, p<0.001]. The main effect of prepulseintensity also achieved significance [F(2,148)=77.66, p<0.001], confirming that increasing prepulse intensity ledsystemically to stronger PPI. However, there was nostatistical evidence in this analysis that this effect of

prepulse intensity was modified by startle intensity (pulseby prepulse interaction: F < 1).

Startle-matched and cross-pulse intensity analysis: percentstartle inhibition

In parallel with the preceding analysis, a supplementaryanalysis on the %PPI measure restricted to the lowest pulse(100 dB) condition when the pulse-alone reaction wascomparable between the two mouse strains was conducted.This effectively circumvented the inappropriateness ofusing %PPI measure under situations of pronounced startledifference between groups. This 2×3 (strain × prepulseintensity) split-plot ANOVA revealed a main effect ofprepulse intensity [F(2,148)=12.24, p<0.001] and itsinteraction with strain [F(2,148)=3.26, p<0.05]. The lattersuggests that DBA/2 PPI deficiency is not dependent on theconfounding reduction in startle reactivity observed at thehigher pulse intensities (Fig. 2b, d).

Next, we conducted the cross-pulse intensity analysison the %PPI measure, as performed above on thereactivity data. To this end, the %PPI data derived fromthe highest pulse intensity (120 dB) data set of the DBA/2mice were compared with that derived from the lowestpulse intensity (100 dB) data set of the C57BL/6 mice. A2×3 (strain × pulse intensity) split-plot ANOVA of this %PPI data set also provided evidence for a relative PPIdeficiency in DBA/2 mice. This analysis yielded asignificant main effect of strain [F(1,74)=6.51, p<0.05],prepulse [F(2,148)=6.51, p<0.001] as well as theirinteraction [F(2,148)=3.43, p<0.05].

Reaction to prepulse-alone trials

Direct response to the prepulse stimulus can be of relevanceto the evaluation of changes in PPI expression (Csomor etal. 2009; Yee and Feldon 2009; Yee et al. 2004a, b, 2005).In comparison with the “no-stimulus” control condition, theprepulse led to a monotonic increase in motor reaction(Fig. 2c). This was largely comparable between the twomouse strains. A 2×4 (strain × prepulse intensity) split-plotANOVA of the ln-transformed mean reactivity scoresyielded only a significant main effect of prepulse intensity[F(3,222)=8.47, p<0.001].

Influence of body weight on startle reaction and PPIexpression

A difference in body weight represents a potential confoundin the measurement of startle response using piezoelectricaccelerometers to detect change in the animal's whole bodymomentum. The body weight of all subjects was thereforerecorded prior to the PPI experiment and was highly


comparable between strains. The mean (±SEM) bodyweight was: C57BL/6=27.6±0.4 g, DBA/2=27.0±0.4 g.A one-way ANOVA failed to reveal a significant straineffect [F(1,74)=1.01, p=0.32]. It is therefore unlikely thatthe divergence between the two strains in PPI expressionand startle reaction could be attributable solely to differ-ences in body weight.

Experiment 2: effects of clozapine on PPI expressionin C57BL/6 and DBA/2 mice

Startle habituation

The finding of a strain-dependent habituation effect wasreplicated, along with overall strain difference in startlereactivity and the effect of pulse intensity. The results aresummarised in Table 1.

In keeping with the outcomes of experiment 1, a 2×2×2×3 (strain × drug × blocks × pulse intensity) split-plotANOVA of the (ln-transformed) startle reactivity obtainedin the first and last blocks of the test session yielded amain effect of blocks [F(1,34)=5.01, p<0.05], pulseintensity [F(2,68)=67.44, p<0.001], strain [F(1,34)=36.95, p<0.001] and the strain × blocks [F(1,34)=36.17,p<0.001] as well as strain × pulse intensity interaction[F(2,64)=28.24, p<0.001]. The directions of all theseeffects were consistent with those observed in experiment1. In addition, clozapine treatment did not affect startlehabituation (drug × block interaction: F < 1), nor did it affectthe overall levels of startle reactivity [drug: F(1,34)=2.07, p=0.16]. However, there was a tendency that clozapinepretreatment weakened reaction to the highest pulsestimulus [drug × pulse intensity: F(2,68)=2.75, p=0.07].This trend is worth noting in view of the followinganalysis.

Startle reaction on pulse-alone trials in the main block

Again, the critical findings concerning strain differenceswere in agreement with the outcomes of experiment 1.Briefly, DBA/2 mice exhibited reduced startle reaction, andthis reduction was more pronounced at the higher pulseintensities (see Fig. 3a, b). A 2×2×3 (strain × drug ×pulse intensity) split-plot ANOVA of pulse-alone trialsderived from the middle block yielded a highlysignificant main effect of strain [F(1,34)=65.71, p<0.001] and its interaction with pulse intensity [F(2,68)=23.55, p<0.001]. In addition, clozapine significantlyaffected startle reactivity, and this drug effect alsodepended on the intensity of the pulse stimulus. Themain effect of drug approached statistical significance[F(1,34)=3.84, p=0.06], and the drug by pulse intensityinteraction was highly significant [F(2,68)=5.38, p<0.01]. T

able

1Sum

maryof

theresults

ofstartle

habituationandprepulse-elicitedreactio

nin

experiment2

DBA/2

mice

C57

BL/6

mice

Clozapine

(n=10

)Saline(n=10

)Clozapine

(n=9)

Saline(n=9)

Startle

habituation

Firstblock

Final

block

Firstblock

Final

block

Firstblock

Final

block

Firstblock

Final

block

100

dB2.6±0.2

1.8±0.2

2.9±0.2

1.9±0.2

2.3±0.2

2.6±0.2

2.6±0.2

2.6±0.2

110

dB2.7±0.3

2.2±0.2

3.5±0.3

2.2±0.2

3.4±0.3

4.1±0.2

3.6±0.3

3.9±0.2

120

dB2.7±0.3

2.1±0.2

3.5±0.3

2.4±0.2

3.8±0.3

4.5±0.3

4.4±0.3

4.8±0.3

Prepu

lse-alon

etrials

65dB

(+0)

1.96

±0.12

2.18

±0.12

2.10

±0.13

2.24

±0.13

71dB

(+6)

2.09

±0.12

2.21

±0.12

2.13

±0.13

2.27

±0.13

77dB

(+12

)2.02

±0.12

2.36

±0.12

2.12

±0.13

2.30

±0.13

83dB

(+18

)2.10

±0.11

2.21

±0.11

2.33

±0.12

2.43

±0.12

The

mean(followinglogarithmictransformation)

reactiv

ityscores

obtained

inthefirstand

lastblocks

ofthetestsessionarepresentedto

allowtheexam

inationof

startle

habituation.

Eachof

these

blocks

consistedof

twotrialsof

each

ofthethreepo

ssiblepu

lse-alon

etrials(100

,110

and12

0dB

A).These

aretabu

latedforeach

ofthefour

strain

×drug

cond

ition

s.Below

that,the

mean(again

follo

winglogarithmic

transformation)

reactiv

ityscores

obtained

inprepulse-alone

trials

arepresentedin

comparisonto

“no-stim

ulus”trials

(i.e.at

theconstant

65-dB

backgrou

ndno

ise).The

standard

errors

(±SE)includ

edwerebasedon

therespectiv

eANOVA

cond

uctedusingSPSSforWindo

ws(version

13).


Similar to the effect of strain, the startle-attenuating effectof clozapine was more notable at higher pulse stimulusintensities, and this drug effect was apparent in bothmouse strains. Pairwise comparisons based on the errorvariance associated with the significant drug by pulseintensity interaction indicate that clozapine pretreatmentled only to a significant reduction in startle reactivity atthe pulse = 120-dB condition.

Prepulse inhibition

Confronted with a significant effect on startle reactivity ofclozapine treatment as well as between strains, analysis ofPPI expression by means of percent inhibition would behighly problematic. Hence, we again adopted the analyticalapproach in experiment 1 and conducted a 2×2×3×4(strain × drug × pulse intensity × prepulse intensity)ANOVA of the ln-transformed reactivity scores obtained

in both pulse-alone and prepulse-plus-pulse conditions (seeFig. 3a, b).

The analysis yielded a main effect of pulse intensity[F(2,68)=93.58, p<0.001], as expected. The overall pres-ence of PPI and its dependence on pulse intensity wereevident from the significant effect of prepulse intensity[F(3,102)=164.05, p<0.001] and its interaction with pulseintensity [F(6,204)=8.37, p<0.001], respectively. A num-ber of significant effects and interactions that emerged areconsistent with strain- and drug-induced differences instartle reactivity as seen already in the previous twoanalyses: strain [F(1,34)=27.97, p<0.001], strain × pulseintensity [F(2,68)=20.46, p<0.001], strain × pulse intensity ×prepulse intensity [F(6,204)=26.74, p<0.001], drug [F(1,34)=6.99, p<0.05] and drug × pulse intensity [F(2,68)=4.82, p<0.05]. As in experiment 1, the interaction between strain andprepulse attained statistical significance [F(3,102)=45.27, p<0.001], suggesting that the reactivity to the pulse stimulus was

Fig. 3 Summary of the results in experiment 2. The average reactivityscores (following logarithmic transformation) across 12 different typesof trial definition obtained in the intermediate block of trials forC57BL/6 mice and DBA/2 mice are depicted in a and b, respectively.Pulse-alone trials are denoted as “+0 dB” in the abscissa. The intensityof the prepulse stimulus in prepulse-plus-pulse trials was also denotedalong the abscissa as +6, +12 and +18 decibel units above the constantbackground noise level of 65 dBA. Within a and b, three separate plotsare illustrated corresponding to the three pulse stimulus intensities:100, 110 and 120 dBA, as indicated at the top. The shaded regions in aand b highlight the respective pulse conditions between the two mouse

strains, which are explicitly compared in all subsequent figures (c–f). cCross-pulse intensity (startle-matched) comparison amongst the fourstrain × drug conditions. d Drug effect derived from this startle-matched comparison by contrasting the two drug conditions (cloza-pine vs. saline) collapsed across strains. e Strain difference collapsedacross drug conditions. f Results expressed in percent inhibition (%PPI) derived from this startle-matched comparison. All mean valuesare illustrated with the associated standard errors (±SE) obtained fromthe respective ANOVA conducted using SPSS for Windows (version13). clz clozapine, sal saline, C57 C57BL/6 strain, DBA DBA/2 strain


less modified by the prepulse stimulus in DBA/2 mice than inC57BL/6 mice (see Fig. 3a, b).

However, the analysis yielded no suggestion thatclozapine treatment had modified the expression of PPIin either strain. Neither the drug by prepulse intensityinteraction [F(3,102)=1.44, p=0.24] nor other higherinteraction terms consisting of these two factors achievedstatistical significance.

We next conducted an additional analysis in an attemptto balance the startle reactivity across different between-subjects conditions. To this end, the data (reactivity scoresobtained from pulse-alone and prepulse-plus-pulse trials)derived from the pulse = 110-dB condition in DBA/2 miceand the data derived from the pulse = 100-dB condition inC57BL/6 mice were combined into a 2×2×4 (strain ×drug × prepulse intensity) ANOVA (see Fig. 3c–e). Asexpected, neither the main effects of strain nor drug weresignificant, and the highly significant main effect ofprepulse intensity [F(3,102)=17.46, p<0.001] indicatedthe presence of PPI. Most critically, this analysis yielded aclear presence of a drug × prepulse intensity [F(3,102)=3.62, p<0.05] and strain × prepulse intensity interaction[F(3,102)=3.07, p<0.05]. These two interaction effectsare depicted separately in Fig. 3d and e, respectively,showing that clozapine exerted an overall PPI-enhancingeffect, whereas DBA/2 mice showed a reduction in PPIexpression relative to C57BL/6 mice. Figure 3c shows thatthe clozapine treatment effect was present in both mousestrains.

Next, this specific startle-matched data set (comprisingdata derived from the lowest pulse condition from C57BL/6mice and the middle pulse condition from DBA/2 mice)was expressed as percent inhibition (%PPI) since the startlereactivity on pulse-alone trials was now largely balanced(see Fig. 3f). The statistical outcomes complementedperfectly the impressions described above, yielding asignificant effect of drug [F(1,34)=9.14, p<0.005] and ofstrain [F(1,34)=11.23, p<0.005]. There was again noindication of any interaction between these two between-subjects factors (F < 1).

Reaction to prepulse-alone trials

Separate analysis of the reaction to the prepulse-alonetrials by a 2×2×4 (strain × drug × prepulse intensity)ANOVA only yielded a significant main effect of prepulseintensity [F(3,102)=3.60, p<0.05]. Neither mouse strainnor clozapine pretreatment was associated with anysignificant change in prepulse-elicited reaction. Theresults are summarised in Table 1, instead of graphicallybecause of the considerable overlap between conditions.The lack of a strain difference in this measure wasconsistent with the outcome from experiment 1. The

absence of a drug effect was instrumental in showing thatit had not altered the direct perception or reaction to theprepulse stimuli.

Experiment 3A: strong latent inhibition of active avoidancelearning (100 pre-exposures of the CS)

Pre-exposure day

The total number of shuttle responses recorded in the pre-exposure phase was subjected to a two-way (strain ×stimulus pre-exposure) ANOVA following a square-roottransformation. As a measure of activity, this yielded asignificant effect of strain [F(1,12)=10.69, p<0.01], withDBA/2 mice performing significantly more spontaneousshuttles than C57BL/6 mice. On the other hand, thismeasure was not affected (F < 1) by whether the animalswere in the PE or nPE condition. The mean (square-root-transformed) total numbers of shuttles recorded in each ofthe four groups were: DBA/2-nPE=15.77, DBA/2-PE=15.18, C57BL/6-nPE=11.26, C57BL/6-PE=12.01 (stan-dard error of differences, SED=1.18).

Conditioning day

Stimulus pre-exposure resulted in an attenuation of avoid-ance learning on the conditioning day, constituting the LIeffect. LI was readily detected in the C57BL/6 mice, butwas distinctly absent in the DBA/2 mice (see Fig. 4a). A2×2×10 (strain × stimulus pre-exposure × 10-trial blocks)split-plot ANOVA of avoidance successes yielded asignificant effect of blocks [F(9,108)=21.01, p<0.001]indicative of a general improvement of avoidance perfor-mance over successive blocks of training. The main effectof pre-exposure approached statistical significance [F(1,12)=4.34, p=0.06], which could be considered assignificant, however, if a one-tailed criterion is adopted,given that expectation of an LI effect has specified thedirection of the pre-exposure effect. Finally, the interactionbetween strain and pre-exposure just reached the margin ofstatistical significance [F(1,12)=4.58, p<0.05]. Pairwisecomparisons confirmed the presence of a significant LIeffect in the C57BL/6 mice (p<0.05), but not in the DBA/2mice. Comparison between strains restricted to either pre-exposure condition did not yield any significant differences,suggesting that the loss of LI in the DBA/2 mice wasattributable to the combined strain differences in both PEand nPE conditions.

The number of spontaneous shuttle responses recordedduring the ITIs on the conditioning day was subjected to a2×2×10 (strain × stimulus pre-exposure × 10-trial blocks)split-plot ANOVA following a square-root transformation.The results indicated that DBA/2 mice were more active


and did not show any sign of reduction over the course ofthe test session as did the C57BL/6 mice (see Fig. 4b).Stimulus pre-exposure did not lead to any change in thisspontaneous activity index. These impressions were con-firmed by the significant effect of strain [F(1,12)=61.07,p<0.001], blocks [F(9,108)=13.25, p<0.001] and theirinteraction [F(9,108)=8.37, p<0.001].

Experiment 3B: weak latent inhibition of active avoidancelearning (50 pre-exposures of the CS)

Pre-exposure day

DBA/2 mice again performed more spontaneous shuttlesthan C57BL/6 mice on the pre-exposure day regardless ofpre-exposure condition. However, in this experiment, thiseffect just failed to achieve statistical significance. A two-

way (strain × stimulus pre-exposure) ANOVA yielded amarginal effect of strain [F(1,12)=4.14, p=0.06]. No othereffects including the interaction term were either close to orachieved statistical significance. The mean (square-root-transformed) numbers of total shuttles recorded in each ofthe four groups were: DBA/nPE=7.97, DBA/PE=9.53,C57/nPE=7.70, C57/PE=7.20 (standard error of differ-ences, SED=0.64).

Conditioning day

As expected, 50 stimulus pre-exposures did not lead to asignificant LI effect in the C57BL/6 mice. Against thisbackground, DBA/2 mice did not show any tendency toexhibit LI. If anything, the direction of the pre-exposureeffect was opposite to that of the LI effect. Overall, DBA/2mice now appeared to show a general impairment in

Fig. 4 Summary of the results in experiments 3A and 3B. Twovariables were obtained on the conditioning day. (1) The number ofavoidance successes expressed as a function of successive ten-trialblocks (on the left) or collapsed across blocks (on the right) forexperiment 3A (a) and Experiment 3B (c), respectively. (2) Thenumber of spontaneous shuttle responses recorded in the inter-trialintervals (square-root-transformed) across successive ten-trial blocksor collapsed across blocks were likewise illustrated in b and d forexperiments 3A and 3B, respectively. Asterisk refers to a significant

(p<0.05) main effect of strain (in b–d) and, in addition, the selectivepresence of a significant pre-exposure effect in the avoidance learningin C57BL/6 mice in experiment 3A (shown in a) and a significant pre-exposure effect in the measure of inter-trial intervals spontaneousshuttles in DBA/2 mice (shown in d). All mean values are illustratedwith the associated standard errors (±SE) obtained from the respectiveANOVA conducted using SPSS for Windows (version 13), which areappropriate for comparison between the illustrated mean values


avoidance learning in comparison to C57BL/6 mice, whichis consistent with the impression obtained in the nPEcondition of experiment 3A. A 2×2×10 (strain × stimuluspre-exposure × 10-trial blocks) split-plot ANOVA ofavoidance successes revealed only a significant maineffect of strain [F(1,12)=12.53, p<0.005], of blocks [F(9,108)=26.76, p<0.001] and their interaction with blocks[F(9,108)=10.78, p<0.001].

Analysis of spontaneous shuttles during ITIs againrevealed a strain difference similar to that seen inexperiment 3A: DBA/2 mice were hyperactive and theirspontaneous shuttle activity failed to show any habituation.The main effect of strain [F(1,12)=41.13, p<0.001], ofblocks [F(9,108)=2.55, p<0.05] and their interaction [F(9,108)=5.82, p<0.001] again emerged as being highlysignificant. In addition, the levels of spontaneous activityexhibited by DBA/2 mice in this experiment appeared alsoto be dependent on the pre-exposure condition—a tendencythat was not present in experiment 3A, when twice as manyCS-pre-exposures had been experienced by the PE animalson the previous day (i.e. in the pre-exposure phase). Thisresulted in a near-significant interaction between strain andpre-exposure [F(1,12)=4.48, p=0.056], and pairwise com-parisons revealed that DBA/PE subjects performed signif-icantly more spontaneous shuttles than subjects in all otherconditions (minimal p<0.02).

Experiment 4: Motor stimulating effects of acuteamphetamine or phencyclidine treatment

Activity in the open field under vehicle treatment conditionconfirmed the impression obtained in experiments 3A and3B that DBA/2 mice were spontaneously more active thanC57BL/6 mice (Fig. 5a). Indeed, the same animals inexperiment 4 had participated in experiments 3A and 3B.Both mouse strains responded to the psychostimulant drugswith an elevation of motor activity. Examination of thetemporal expression of the drug effect suggest that DBA/2mice responded more quickly to amphetamine, achievingpeak response almost within the first 10 min, whilst it tookC57BL/6 mice nearly 30 min to reach peak response(Fig. 5b). The peak response magnitude appeared compa-rable between strains, and by the end of the 2-h observationperiod, activity between the two strains was comparable,but was still higher than their respective vehicle-treatedcontrols. In contrast, the immediate response to acute PCPinjection was highly comparable between DBA/2 andC57BL/6 mice. However, DBA/2 mice showed a sustainedand elevated (relative to DBA/saline subjects) responsethroughout the 2-h period, whereas activity levels inC57BL/6 mice showed a consistent (except from bins 8 to9) monotonic reduction following the immediate peak(Fig. 5c). Hence, although DBA/2 mice showed some form

of hyperresponsiveness to amphetamine and PCP incomparison to C57BL/6 mice, the precise temporal dynam-ic of this strain difference differed substantially between thetwo pharmacologically distinct drugs.

The interpretations above were substantiated by theresults of a 2×3×12 (strain × drug × 10-min bins) split-plot ANOVA of distance travelled (square-root-trans-formed) per bin. All main effects attained statisticalsignificance: strain [F(1,26)=34.73, p<0.001], drug [F(2,26)=74.64, p<0.001] and bins [F(11,286)=15.00, p=0.001]. The presence of a strain by drug interaction [F(2,26)=6.85, p<0.005] and the three-way interaction [F(22,286)=3.90, p=0.001] confirmed the differences be-tween strains in terms of their overall activity levels, andthat the temporal profiles differed between the three drugconditions. Because the drug-induced activity profilediffered between the two drugs (and in comparison tosaline vehicle), the interaction between drug and bins wasalso highly significant [F(22,286)=7.87, p<0.001].

Pairwise comparisons between strains under each treat-ment condition confirmed the impression that the twostrains differed under vehicle condition (p<0.001) andfollowing PCP treatment (p<0.001), but not when theywere both under the influence of amphetamine (p=0.60).The latter supported the interpretation that the two strainsdiffered primarily in their temporal response to amphet-amine. This emphasis was also confirmed by supplemen-tary restricted analyses: A significant strain by binsinteraction was only evident in the analyses restricted eitherto the amphetamine [F(11,110)=2.29, p<0.05] or PCP [F(11,110)=8.03, p<0.001] conditions, whereas the straindifference was stable over time in the absence of a drugchallenge (vehicle condition).

Discussion

The present study represents a comprehensive evaluation ofthe DBA/2 mouse strain in schizophrenia-related endophe-notypes and its reaction to acute psychostimulant drugchallenge in comparison to the C57BL/6 strain. This straincomparison has revealed a number of behavioural diver-gences that may be interpreted as the presence of multipleschizophrenia-related traits in DBA/2 mice (or, alternative-ly, of an antipsychotic profile in C57BL/6 mice). Thisincludes, first of all, the relative deficiency in prepulseinhibition and latent inhibition, which parallel that reportedin schizophrenia patients (e.g. Baruch et al. 1988; Braff etal. 2001, 1992; Gray et al. 1992). The difference in motorresponse to acute psychostimulant drug challenge furtherindicates that dopaminergic as well as glutamatergic/NMDA receptor functions also diverge between these twomouse strains. It is, however, premature to conclude that


these presumed differences in neurotransmission functionare causally related to the behavioural divergence observed.In addition to confirming earlier speculations on thebehavioural and cognitive characteristics of DBA mice(e.g. Baarendse et al. 2008; Olivier et al. 2001), the presentstudy also provides fresh insights that are highly relevant tothe purported use of this strain comparison as a means tomodel selected schizophrenia traits in animals in terms offace, predictive (of drugs with antipsychotic potential) andconstruct (e.g. genetic, developmental and physiologicaldifferences) validity.

Deficient prepulse inhibition against a backgroundof reduced startle reactivity in DBA/2 mice

We have first addressed the reliability of the reportedrelative PPI deficiency in DBA/2 mice because it is knownthat confounding change in startle reaction has beenrepeatedly noted in previous reports (e.g. Bortolato et al.2007; Crawley et al. 1997; McCaughran et al. 1997; Paylorand Crawley 1997). Ignoring such confounding differencesin startle reactivity can seriously undermine the validity ofany claim as to whether PPI expression differs betweengroups (see Swerdlow et al. 2000). As can be readily seenin Figs. 2a and 3b, the absolute reduction in startlereactivity was heavily influenced by the correspondingbaseline reactivity to the startle-eliciting pulse stimulus inC57BL/6 mice: This sensitivity to the prepulse stimulus

was less pronounced when the startle reaction to the pulsestimulus was weaker. Hence, it is not inconceivable that theapparent deficiency in PPI observed in DBA/2 mice wasrelated to their generally weaker startle reactivity. Ittherefore poses an important interpretative problem becausethis confound might lead to the identification of startle-corrective therapy rather than PPI-corrective treatment. Toovercome this confounding effect, we adopted a recentlydeveloped PPI test paradigm incorporating multiple pulsestimulus intensities (Yee et al. 2005), which allowed thestartle reaction between the two strains to be balanced oreven reversed, thus providing an insight into a possibledependency on the existing strain difference in startlereactivity. The principle of its application is similar forboth experiments 1 and 2, although the critical (crossedpulse intensity) comparisons being emphasised differedsomewhat. The conclusions were identical: The relative PPIdeficiency in DBA/2 mice cannot be solely attributed totheir weaker startle response. The robustness and flexibilityof this analytical approach in overcoming baseline startlereactivity difference in general was thus also demonstrated(Csomor et al. 2008; Yee et al. 2005). Based on thisapproach, we can also conclude that DBA/2 mice cannot becharacterised as “low-startling” C57BL/6 mice in terms ofthe PPI expression. We have previously shown that when arandom sample of C57BL/6 mice were segregated intosubgroups differing in startle reactivity, and the subgroupsthen compared, the apparent differences in PPI indexes

Fig. 5 Summary of the results of experiment 4. Distance travelled per10-min bins following vehicle saline (a), amphetamine (b) andphencyclidine (PCP) (c) is separately plotted to contrast the twomouse strains (C57 C57BL/6, DBA DBA/2). Insets in each graphrepresent the mean distance travelled per bin collapsed across bins toillustrate the relative augmentation of locomotor response in DBA/2

mice (in a and c, as indicated by asterisk). All mean values areillustrated with the associated standard errors (±SE) obtained from theoverall three-way ANOVA conducted using SPSS for Windows(version 13), which are appropriate for comparisons between theillustrated mean values


calculated according to convention (%PPI) would beeliminated when a crossed pulse intensity comparison wasconducted to balance the reaction on pulse-alone conditions(Yee et al. 2005). This method of analysis thus showed thatPPI expression was essentially comparable betweenC57BL/6 mice differing in startle magnitude. There is noreason to suspect that the same situation would not hold forother inbred mouse strains. Hence, the confirmation of thestrain difference in PPI expression in the present study hasclarified considerably the concern over the confoundingdifference in startle reactivity. We conclude that theregulation of PPI expression was different between thesetwo mouse strains.

A schizophrenia-related trait in C57BL/6 in comparisonto DBA/2 mice?

Experiment 1 showed clearly the absence of startlehabituation in C57BL/6 mice against the clear presence ofstartle habituation in the DBA/2 mice (Fig. 2a). Aconsistent pattern of results was replicated in experiment2 (Table 1). Indeed, the lack of startle habituation inC57BL/6 mice is a robust result in agreement with earlierreports (Pietropaolo et al. 2008; Yee et al. 2004a, 2005).Given that impairment in startle reflex habituation is alsopresent in schizophrenia patients and may be related toimpaired selective attention (Geyer and Braff 1987; Geyeret al. 1990), should one also consider this strain-specifictrait as a potential schizophrenia-related endophenotype?This certainly warrants further assessment, including thepsychological nature of this phenotype as well as itsrelationship to PPI disruption in patients. Our data providethree relevant perspectives in this regard: (1) Clozapine wasineffective in promoting startle habituation in C57BL/6mice (see Table 1), although the mice were responsive tothe drug in terms of PPI expression (experiment 2). (2) ThePPI deficiency demonstrated in DBA/2 mice relative toC57BL/6 mice was obviously not associated with animpairment in startle habituation (experiments 1 and 2).(3) The startle habituation deficit seen in C57BL/6 micecannot be generalised to other responses because habitua-tion of locomotor activity in the open field was equallyobserved in C57BL/6 and DBA/2 mice (Fig. 4a). Hence, itis unlikely that C57BL/6 mice suffer from a generalisedweakening of habituation and/or strengthening of sensitisa-tion processes.

It has not escaped our notice that the presence of startlehabituation in DBA/2 but not C57BL/6 mice might havecontributed to the overall strain difference in startlereactivity seen across the test session (as recorded in themiddle block of the test session used to assess PPI). Wetherefore explicitly compared all PPI-related measures bysplitting the session into two halves and repeated all

relevant analyses. This did not yield any deviation fromour overall conclusion described above. Therefore, thedifference in startle habituation does not undermine theconfidence in our interpretations of the PPI data. Our resultspoint to an intriguing double dissociation1 in this straincomparison, with deficiency in one form of startle plasticityfound only in one mouse strain (PPI deficiency in DBA/2)and another form of anomaly only in the other strain (startlehabituation deficit in C57BL/6). Interestingly, both pheno-types are of relevance to schizophrenia. This doubledissociation in itself calls for caution in the use, orinterpretation, of a given pairwise strain comparison as ageneral model of schizophrenia because either strain maybe considered as more or less schizophrenia-like. Thecomparison should be made in a trait-by-trait, orendophenotype-by-endophenotype, manner (Arguello andGogos 2006).

Responsiveness to systemic clozapine in PPI expression

Experiment 2 examined whether the baseline difference inPPI expression which emerged between DBA/2 andC57BL/6 mice would suggest that the former strain mighttherefore be more powerful in detecting the effect ofatypical antipsychotic drugs, as exemplified by clozapine.If so, then one may conclude that the DBA/2 strain mayrepresent a disease model such that its sensitivity toantipsychotic drugs signifies a correction or normalisationof PPI function, at least within the restrictive comparison toC57BL/6 mice. Thus, the answer to this question wouldbear significance for the increasingly widespread use ofDBA/2 mice (often exclusively, without a comparisonstrain) as a model to detect the PPI-corrective property ofnovel antipsychotic drugs (e.g. Boulay et al. 2008;Depoortère et al. 2005; Fox et al. 2005; Kinney et al.2003), as if another strain would yield a false negativeoutcome. The results of experiment 2 are clear: Against ademonstrated baseline (vehicle treatment) strain differencein PPI expression (Fig. 3e), clozapine was efficacious insignificantly enhancing PPI in both mouse strains (Fig. 3d).These findings therefore do not support the suggestion thatDBA/2 mice are superior (at least to C57BL/6 mice) indetecting the effect of atypical antipsychotic drugs (cloza-pine here) on PPI. This conclusion is consistent with anearlier study by McCaughran et al. (1997) who showed thatthe C57BL/6 and DBA/2 were similarly effective indetecting the PPI-enhancing effect of clozapine as well as

1 Strictly speaking, this is not a classical double dissociation due to alack of a common control comparison group against which the DBA/2and C57BL/6 strains can be compared. Here, DBA/2 mice showed arelative impairment in PPI, whilst C57BL/6 mice exhibited a lack ofstartle habituation when they were compared with each other.


other atypical and typical antipsychotic drugs. However, adose–response analysis would be necessary to consolidatesuch a claim by showing that when startle reactivity isbalanced between strains, the dose–response curves of thetwo strains would overlap each other. Our present experi-ment had investigated a single dose of clozapine only, andthe conclusion therefore needs to be restricted to thatparticular dose. Nonetheless, the responsiveness of C57BL/6 mice to this dose was clear and comparable to that seen inDBA/2 mice.

Again, the crossed pulse intensity analysis has beenhighly instrumental in allowing us to arrive at this criticalconclusion amidst the confounding effects on startlereactivity (pulse-alone trials) of clozapine treatment as wellas between mouse strains. It also suggests that at the highestpulse intensity of 120 dB, clozapine merely produced anoverall downward shift of startle reactivity in both mousestrains (Fig. 3a, b). In such case, expression of PPI aspercent startle inhibition relative to pulse-alone reactivitywould give an erroneous impression of PPI enhancement.This is most obviously seen in the report by Ouagazzal etal. (2001) who claimed that clozapine was effective inenhancing PPI, based on %PPI analysis alone, at a dose of30 mg/kg when startle response was massively reduced toless than one fifth of the control group level (i.e. a 82%reduction in startle reactivity) due presumably to thesedative effect of clozapine at such high doses. Our analysisovercame this problem and further indicated that the PPI-enhancing effect of clozapine was more readily shown inboth strains at lower pulse intensities (100–110 dB). Thismay suggest that if DBA/2 mice are more sensitive to thiseffect of clozapine (and by extension other atypicalantipsychotic drugs), it is because they inherently exhibitlower startle reactivity. This, however, can be readilymimicked by the use of lower pulse intensity in C57BL/6mice, and under such conditions, the PPI-enhancing effectof clozapine can also be readily detected. Hence, whilst weshowed that the PPI deficiency of DBA/2 relative toC57BL/6 mice cannot be explained solely due to theconfounding change in startle response (experiment 1), thereported impression that DBA/2 mice seem to be moresuitable for the detection of atypical antipsychotic propertyin PPI may indeed stem from the reduced startle reactionnative to this mouse strain. This highlights anotherimportant potential benefit of PPI tests that include multiplepulse stimulus intensities.

Another possibility is that DBA/2 mice were moreresponsive than C57BL/6 to disruption by NMDA receptorantagonists and thereby more suitable or sensitive indetecting atypical antipsychotic drugs (Geyer 2006a, b).First of all, PPI expression in C57BL/6 is indeed sensitiveto NMDA receptor antagonism (Yee et al. 2004a), so thecomparison must be based on relative rather than absolute

sensitivity. Using a dose–response analysis (0, 10, 15 or20 mg/kg), Spielewoy and Markou (2004) compared thesensitivity to PCP in four mouse strains (C57BL/6, DBA2,C3H/He and 129T2/SvEms) and concluded that C57BL/6was the least sensitive, or even statistically insensitive.However, it seems that these authors have overinterpretedtheir own data since the conclusion was inconsistent acrossexperiments (see their Figs. 1 and 3). Closer examination oftheir data also suggests that the dose–response relationshipin C57BL/6 mice was somewhat shifted to the left of that ofDBA/2 mice (see their Fig. 1a, b). Finally, their PPIevaluation failed to take into account the demonstratedstrain-dependent effects of PCP treatment on startlereactivity that would certainly confound the interpretationof the data (see their Fig. 2). We therefore conclude thatthere is no convincing evidence to date to suggest thatDBA/2 and C57BL/6 mice differ substantially in theirresponsiveness to NMDA antagonists-induced PPI disrup-tion. This issue should be addressed directly, but cautionover confounding changes in startle reactivity must beexercised.

Beyond PPI deficiency

DBA/2 mice exhibited a parallel deficiency in the expres-sion of LI under conditions in which C57BL/6 miceshowed robust LI (Experiment 3a, Fig. 4a), therebyextending the recent finding by Baarendse et al. (2008)who compared LI expression between DBA/2 and C57BL/mice in a passive avoidance paradigm. Similar to PPI, LI isalso impaired in at least a subset of schizophrenia patients(e.g. Baruch et al. 1988; Gray et al. 1992). Whilst PPI issuggested to measure the pre-attentive filtering or gatingprocess (e.g. Ellenbroek 2004; Geyer 2006a), LI is moreclosely related to learned inattention, i.e. learning to ignoreirrelevant stimuli in one’s environment (e.g. Gray et al.1991, 1995; Lubow 1989; Mackintosh 1973; Wagner1978). The concomitant presence of PPI and LI deficiency(in DBA/2 mice relative to C57BL/6 mice) demonstratedhere in a within-subject manner lends further evidence thatthis strain comparison may capture more than one endo-phenotype related to schizophrenia.

The particular form of LI disruption in DBA/2 micecame about through a loss of the stimulus pre-exposureeffect, i.e. CS pre-exposure failed to slow down subsequentlearning about the predictive value of that stimulus, as wellas through an impairing effect on learning in the non-pre-exposed condition—a novel finding that was mostclearly shown in experiment 3B (Fig. 4c). It has beensuggested that the latter effect might be interpreted as: (1) ageneral learning deficit—the DBA/2 mice learned lessabout the CS–US association, just as the DBA/2 PE micehad learned less about the CS–nothing association or the


irrelevance of the CS during pre-exposure (see Yee et al.1995), or (2) a failure to benefit from the context pre-exposure effect (i.e. latent inhibition to the context (e.g.Killcross et al. 1998), which should reduce the salience ofthe context and enhance attention to the CS (and therebylearning of its predictive significance) during subsequentconditioning for the NPE animals (see Pouzet et al. 2004).It should also be emphasised that the relative deficit inavoidance learning in DBA/2 mice was expressed againstthe background of enhanced spontaneous shuttle responsesrecorded in the ITIs, with C57BL/6 mice exhibiting verylow levels of spontaneous shuttles from the second blockonwards. Hence, the poorer performance in DBA/2 micecannot be explained by the relative difference in spontane-ous shuttles. The higher spontaneous shuttle rate in DBA/2mice would, if anything, allow some successful avoidanceresponses to be made by chance alone. The conclusion thatDBA/2 mice are impaired in at least this form of associativelearning is a novel observation that certainly warrantsfurther evaluation. This is in direct conflict with twoprevious studies showing an opposite strain difference in asimilar two-way active avoidance task (Bovet et al. 1969;Vetulani et al. 1989) when training was extended beyondthe first hundred trials. Although we had only performed100 trials here, it is unlikely that our observed pattern ofresults would reverse if additional training was conducted.A survey on the existing literature, however, does not seemto yield a clear consensus on this particular strain differencein avoidance learning, with some reports also failing toidentify any clear performance difference (see Bovet et al.1969; Iso and Shimai 1991; Sprott and Stavnes 1975;Stavnes and Sprott 1975b; Vetulani et al. 1989; Wahlsten1972; Weinberger et al. 1992). A number of factors havebeen proposed to account for such apparent inconsistency,including shock intensity (Carran et al. 1964; Wahlsten1972), the modality of the conditioned stimulus (Oliverio1967), the training protocol (Wimer et al. 1968) and the ageof the animals (Stavnes and Sprott 1975a). Anotherpossibility is the precise source or stock of the miceobtained. Genetic drift away from the original Jacksonstocks may occur to varying degrees depending on thebreeding system maintained by the supplying laboratory.Unfortunately, no single factor can yet fully account forall reported results.

With respect to the precise pattern of LI abolition shownin experiment 3A, it closely resembles that seen followingcell lesion of the entorhinal cortex (Yee et al. 1995, 1997),which abolishes LI via a disruption of the entorhinal–striatal projection, thereby disturbing the balance betweenglutamatergic and dopaminergic activities in the nucleusaccumbens (Gray et al. 1995). On the other hand, it isdistinctly different from the effect of cell lesions of thehippocampus, which lead to abnormally persistent LI

(Honey and Good 1993). It therefore contradicts thedescription of the DBA/2 strain (in comparison of theC57BL/6 strain) as a model or manifestation of a loss ofhippocampal function, with relatively poor performance(compared with C57BL/6 mice) in tasks requiring theformation of complex contextual representations includingspatial learning as well as contextual fear conditioning (e.g.Ammassari-Teule et al. 2000a, 2000b, 1995; Barber et al.1974; Gerlai 1998; Paylor et al. 1993, 1994; Wehner et al.1990; Wimer et al. 1976).

The aberrant persistence of LI expression has beenemphasised by Weiner (2003) as a model of negative andcognitive symptoms of schizophrenia. It can be induced inanimals by NMDA receptor blockade or selective brainlesions to the nucleus accumbens core, basolateral amyg-dala or orbitofrontal cortex, and its normalisation has beenspecifically linked to atypical (but not typical) antipsy-chotic drugs (Gaisler-Salomon and Weiner 2003; Gaisler-Salomon et al. 2008; Lipina et al. 2005; Schiller et al.2006). Experiment 3B examined this possibility by using asmaller number of stimulus pre-exposures, which wasinsufficient to generate LI in C57BL/6 mice (and thesensitivity of this procedure to the induction of abnormal-ly persistent LI by NMDA receptor blockade has beenconfirmed in our laboratory; W-N. Zhang, P. Singer, J.Feldon, and B.K. Yee, unpublished data). Against thisbackground, LI was similarly absent in the DBA/2 mice,which also exhibited again an overall deficit in avoidancelearning. Extension of this null finding to other proceduralmanipulations designed to minimise LI expression in thereference control group (e.g. increased number of CS–USpairings or contextual change between pre-exposure andconditioning) would be highly relevant. One previousstudy has reported that 16 CS pre-exposures (distributedacross 4 days) followed by two CS–US pairings in aconditioned freezing paradigm led to LI in DBA/2 but notC57BL/6 mice (Restivo et al. 2002), but the statisticalevidence provided was weak, and careful examinationrevealed that their test condition was far from ideal—witha trend of LI effect obtained in C57BL/6 mice, reportedly atp=0.06.

According to the “two-headed” latent inhibition modelof schizophrenia formulated by Weiner (2003), our LIexperiments with DBA/2 mice would suggest an interpre-tation of this mouse strain in terms of positive symptomsrather than negative/cognitive symptoms of schizophrenia.The latter in particular would disagree with the suggestionthat the PPI relative deficiency demonstrated in experi-ments 1 and 2 would be predictive of cognitive symptomsin schizophrenia (Geyer 2006a). As discussed below, theoutcomes of the psychostimulant drugs experiment haveyielded some support to both perspectives in terms ofneuropharmacology.


Altered locomotor response to dopamine receptor agonismand NMDA receptor antagonism in DBA/2 mice

One physiological perspective on the distinction betweenpositive and negative/cognitive symptoms of schizophreniaemphasises the respective contributions of dopaminergichyperfunction and glutamatergic/NMDA receptor hypo-function. As mentioned above, these two hypothesisedpathophysiological mechanisms are both highly relevant tothe interpretation and application of the PPI and LI modelsof the relevant psychopathology. The LI phenotypes ofDBA/2 mice here are suggestive of dopaminergic dysfunc-tion (Weiner 2003), whereas the PPI deficiency and itsresponsiveness to clozapine may be indicative of glutama-tergic/NMDA receptor hypofunction (Geyer 2006a, b).These suggestions were therefore examined by acutesystemic challenge of a dopamine-releasing drug (amphet-amine) or an NMDA receptor blocker (PCP) in theseanimals (experiment 4).

First of all, a clear baseline difference was observed withDBA/2 mice showing a higher level of activity (Fig. 5a),which was also seen on the pre-exposure day of the LIexperiments with a different activity measure, namely totalnumber of spontaneous shuttles (experiments 3A and 3B).We have since also replicated this finding in a separatecohort of naïve DBA/2 and C57BL/6 mice, and thus, priortest experience here does not seem to account for thepresent outcome (P. Singer, J. Feldon and B.K. Yee,unpublished observation). This impression is in keepingwith some (Morse et al. 1993; Tohmi et al. 2005) but notother reports that have either shown an opposite pattern ofresults (e.g. Cabib et al. 2000, 2002; McNamara et al. 2006)or failed to reveal any difference between these two mousestrains (Cabib et al. 2000; Ventura et al. 2004). Suchinconsistency may stem from the critical importance ofenvironmental factors including housing conditions (Morseet al. 1993), feeding regime (Cabib and Bonaventura 1997)and light/dark phase of the diurnal cycle (P. Singer, J.Feldon and B.K. Yee, unpublished observation). Suchenvironmental factors in conjunction with the reportedenhanced sensitivity to stress in DBA/2 mice in comparisonto C57BL/6 mice (see e.g. Badiani et al. 1992; Cabib et al.2000) may also contribute to the observed results in thedrug phase of the experiment here.

Although the baseline (vehicle treatment) differencebetween strains renders interpretation of the overallmagnitude of drug-induced locomotor response somewhatdifficult, the temporal profile of the response clearlydiffered between the two mouse strains. The onset andoffset of the peak response to amphetamine were acceler-ated in DBA/2 mice by 10–20 min relative to C56BL/6mice (Fig. 5b), with little change in the overall absoluteactivity levels. This outcome is in conflict with previous

studies reporting a relative reduction of motor responsein DBA/2 mice that has been linked to an increase inamphetamine-induced dopamine release in the prefrontalcortex, which in turn suppresses dopamine release inthe nucleus accumbens that mediates the hyperactivityeffect (Ventura et al. 2004; Zocchi et al. 1998). However,when the baseline difference in activity level observedhere is taken into account, the motor stimulant effectof amphetamine might, on the contrary, be consideredweaker in the DBA/2 mice relative to the C57BL/6 miceand would not be in direct disagreement with the reportsby Ventura et al. (2004) and Zocchi et al. (1998).Nonetheless, the more rapid onset of hyperlocomotorresponse to amphetamine in the DBA/2 mice here mightbe expected on the basis of known dopaminergic differ-ences between DBA/2 and C57BL/6 mice. Firstly, arelative reduction in D2-like dopamine receptors density inthe nucleus accumbens (Puglisi-Allegra and Cabib 1997)might allow a higher proportion of the receptors to beactivated more quickly in DBA/2 mice. Secondly,enhanced dopamine release in the medial prefrontalcortex (Ventura et al. 2004) and a higher number ofD2-like receptors in the ventral tegmental area andsubstantia nigra, in combination with a higher dopami-nergic autoreceptors-to-postsynaptic receptors ratio in thestriatum (Puglisi-Allegra and Cabib 1997), might allcontribute to facilitate the inhibition of mesolimbic dopa-minergic hyperactivity induced by amphetamine (Zocchiet al. 1998).

In response to PCP, both mouse strains showed anincrease in activity to a similar absolute magnitude withinthe first bin (Fig. 5c), which is in contrast to the initialimpact of amphetamine challenge such that the baselinestrain difference persisted in the first two bins post-injection(Fig. 5b). Subsequently, the activity level in the PCP-treatedC57BL/6 mice gradually returned to that of vehicle-treatedC57BL/6 mice. In contrast, the cessation of the motorresponse to PCP challenge was less rapid in the DBA/2mice and still had not returned to control levels by thesession’s end. Hence, even if one argues that their initialresponse was weaker than C57BL/6 mice in terms ofproportional elevation of activity level relative to vehicletreatment, the interpretation that their response was pro-tracted in time remains non-negotiable. One way toovercome the confounding difference in baseline activitydifference (pre-drug phase in a within-subject design orvehicle treatment as in a between-subject design adoptedhere) is to use a subtraction scoring method. Based on this,Alexander et al. (1996) concluded that DBA/2 and C57BL/6 mice do not differ in their overall response to PCP,although they did not provide a temporal profile analysis.Given that current data on this issue are scant, furtherinvestigations are certainly warranted.


The absence of concurrent neurochemical measures inthe present study, however, precludes an extended discus-sion of the possible differences in neurotransmitter func-tions between DBA/2 and C57BL/6 mice. Nonetheless, theevidence here is suggestive of differences between thesetwo mouse strains in both dopaminergic and glutamatergicfunctions, including perhaps their interaction. It remains tobe determined whether these neuropharmacological differ-ences may underlie some of the key behavioural andcognitive divergence between them.

A postscript on the caveats in strain comparison studies

Whilst strain comparison between inbred strains of labora-tory rodents can be a fruitful approach in behaviouralgenetics, the present study also illustrates that care andcaution must be exercised in the interpretation of the databecause it is often the case that different strains areassociated with multiple behavioural differences. Thesecan be significant confounding variables in one particularexperiment, yet not in another. The present study wasmotivated first and foremost by the need to address thesignificant confounding changes in startle reactivity andtheir possible impact on interpreting PPI data. This wasexplicitly addressed here for the first time and was resolvedusing a novel paradigm designed particularly for dealingwith such a confound. In addition, critical confoundingchanges were also recognised in all subsequent testsconducted. In the assessment of LI, the results wereconfounded by strain difference in avoidance learning assuch, which in turn was confounded by differences in thefrequency of spontaneous shuttle responses. In both experi-ments (experiments 3A and 3B), appreciation of theseconfounds have enriched rather than limited our interpre-tation of the data. In evaluating the motor stimulating effectof psychomimetic drugs, interpretation of the data derivedfrom the drugged groups was confounded by baseline straindifference in the vehicle treatment condition. This, howev-er, cannot be easily overcome, except by the use of awithin-subject design in combination with prolongedapparatus habituation to minimise or eliminate suchbaseline differences.

Conclusion

There is a strong genetic component to schizophrenia, andtherefore, comparative studies of inbred laboratory rodentsoffer an opportunity to identify the genetic and associatedphysiological mechanisms that may underlie selectedendophenotypes of the disease. The comparison betweenDBA/2 and C57BL/6 mice may represent a promisingcomparative couple perhaps in the search of potential

candidate genes, neural mechanisms and neurodevelopmen-tal divergences that may underlie the pathophysiology ofmultiple schizophrenia-related endophenotypes. It is awelcomed addition to the bottom-up approach of pointmutation generated by genetic engineering. However,unlike treatment-based disease models, such as mutations,drugs or specific neurodevelopmental interference, theprecise cause and effect relationship underlying straindifferences is particularly difficult to pin down becausethe mouse strains are “given to” rather than “by design of”the experimenter. To harness the power of such mousestrain models, testable hypotheses targeting individualphenotypic components ought to be thoroughly evaluatedby other experimental approaches.

Lastly, it is essential to recognise that comparativeanalysis of behavioural phenotypes between strains isnecessarily relative in nature and is heavily influenced bythe choice of the control or reference strain. The behav-ioural deficiencies in the DBA/2 mice revealed here are allbased on a unique comparison with the C57BL/6 strain, andit is not at all difficult to conceive that a very differentpicture may emerge if another reference strain wereemployed. It would be inappropriate to refer to theDBA/2 mouse strain as a “psychotic” mouse strain inabsolute terms. In contrast, schizophrenia is recognised as adisease entity in the clinic, and the diagnostic criteria aredesigned to minimise such arbitrary relativism. Relativelyspeaking, the DBA/2 strain may be associated with apropensity to develop behavioural traits that may resemblethose of schizophrenia, but this does not imply that thesemice are suffering from psychosis.

Acknowledgments The present study was supported by the FederalInstitute of Technology Zurich and the National Centre of Competencein Research (NCCR): Neural Plasticity and Repair funded by theSwiss National Science Foundation.

References

Alexander RC, Wright R, Freed W (1996) Quantitative trait locicontributing to phencyclidine-induced and amphetamine-inducedlocomotor behavior in inbred mice. Neuropsychopharmacology15:484–490

Ammassari-Teule M, Passino E, Restivo L, de Marsanich B (2000a)Fear conditioning in C57BL/6 and DBA/2: variability in nucleusaccumbens according to the strain predisposition to showcontextual- or cue-based responding. Eur J Neurosci 12:4467–4474

Ammassari-Teule M, Restivo L, Passino E (2000b) Contextual-dependent effects of nucleus accumbens lesions on spatiallearning in mice. Neuroreport 11:2485–2490

Ammassari-Teule M, Tozzi A, Rossi-Arnaud C, Save E, Thinus-BlancC (1995) Reactions to spatial and nonspatial change in twoinbred strains of mice: further evidence supporting the hippo-


campal dysfunction hypothesis in the DBA/2 strain. Psychobiol-ogy 23:284–289

Arguello PA, Gogos JA (2006) Modeling madness in mice: one pieceat a time. Neuron 52:179–196

Badiani A, Cabib S, Puglisi-Allegra S (1992) Chronic stress inducesstrain-dependent sensitization to the behavioral effects ofamphetamine in the mouse. Pharmacol Biochem Behav 43:53–60

Baarendse PJ, van Grootheest G, Jansen RF, Pieneman AW, OgrenSO, Verhage M, Stiedl O (2008) Differential involvement of thedorsal hippocampus in passive avoidance in C57bl/6J and DBA/2J mice. Hippocampus 18:11–19

Barber RP, Vaughn JE, Wimer RE, Wimer CC (1974) Genetically-associated variations in the distribution of dentate granule cellssynapses upon the pyramidal cell dendrites in the mousehippocampus. J Comp Neurol 156:417–434

Baruch I, Hemsley D, Gray JA (1988) Differential performance ofacute and chronic schizophrenic in a latent inhibition task. J NervMent Dis 176:598–606

Bortolato M, Frau R, Orrù M, Piras AP, Fà M, Tuveri A, PulighedduM, Gessa GL, Castelli MP, Mereu G, Marrosu F (2007)Activation of GABA(B) receptors reverses spontaneous gatingdeficits in juvenile DBA/2J mice. Psychopharmacology 194:361–369

Boulay D, Pichat P, Dargazanli G, Estenne-Bouhtou G, Terranova JP,Rogacki N, Stemmelin J, Coste A, Lanneau C, Desvignes C,Cohen C, Alonso R, Vigé X, Biton B, Steinberg R, Sevrin M,Oury-Donat F, George P, Bergis O, Griebel G, Avenet P, ScattonB (2008) Characterization of SSR103800, a selective inhibitor ofthe glycine transporter-1 in models predictive of therapeuticactivity in schizophrenia. Pharmacol Biochem Behav 91:47–58

Bovet D, Bovet-Nitti F, Oliverio A (1969) Genetic aspects of learningand memory in mice. Science 163:139–149

Braff DL, Geyer MA, Swerdlow NR (2001) Human studies of prepulseinhibition of startle: normal subjects, patient groups, and pharma-cological studies. Psychopharmacology (Berl) 156:234–258

Braff DL, Grillon C, Geyer MA (1992) Gating and habituation of thestartle reflex in schizophrenic patients. Arch Gen Psychiatry49:206–215

Browman KE, Komater VA, Curzon P, Rueter LE, Hancock AA,Decker MW, Fox GB (2004) Enhancement of prepulse inhibitionof startle in mice by the H3 receptor antagonists thioperamideand ciproxifan. Behav Brain Res 153:69–76

Bullock AE, Slobe BS, Vazques V, Collins AC (1997) Inbred mousestrains differ in the regulation of startle and prepulse inhibition ofthe startle response. Behav Neurosci 111:1353–1360

Cabib S, Bonaventura N (1997) Parallel strain-dependent susceptibil-ity to environmentally-induced stereotypies and stress-inducedbehavioral sensitization in mice. Physiol Behav 61:499–506

Cabib S, Orsini C, Le Moal M, Piazza PV (2000) Abolition andreversal of strain differences in behavioral responses to drugs ofabuse after a brief experience. Science 289:463–465

Coyle JT (2006) Glutamate and schizophrenia: beyond the dopaminehypothesis. Cell Mol Neurobiol 26:365–384

Cabib S, Puglisi-Allegra S, Ventura R (2002) The contribution ofcomparative studies in inbred strains of mice to the understand-ing of the hyperactive phenotype. Behav Brain Res 130:103–109

Carlsson A (1988) The current status of the dopamine hypothesis ofschizophrenia. Neuropsychopharmacology 1:179–186

Carran AB, Yeudall LT, Royce JR (1964) Voltage level and skinresistance in avoidance conditioning in inbred strains of mice. JComp Physiol Psychol 58:427–430

Chang T, Meyer U, Feldon J, Yee BK (2007) Disruption of the USpre-exposure effect and latent inhibition in two-way activeavoidance by systemic amphetamine in C57BL/6 mice. Psycho-pharmacology (Berl) 191:211–221

Crawley JN, Belknap JK, Collins A, Crabbe JC, Frankel W,Henderson N, Hitzemann RJ, Maxson SC, Miner LL, Silva AJ,Wehner JM, Wynshaw-Boris A, Paylor R (1997) Behavioralphenotypes of inbred mouse strains: implications and recom-mendations for molecular studies. Psychopharmacology (Berl)132:107–124

Csomor PA, Yee BK, Quednow BB, Stadler RR, Feldon J,Vollenweider FX (2006) The monotonic dependency of prepulseinhibition of the acoustic startle reflex on the intensity of thestartle-eliciting stimulus. Behav Brain Res 174:143–150

Csomor PA, Yee BK, Feldon J, Theodoridou A, Studerus E,Vollenweider FX (2009) Impaired prepulse inhibition andprepulse-elicited reactivity but intact reflex circuit excitability inunmedicated schizophrenia patients: a comparison with healthysubjects and medicated schizophrenia patients. Schizophr Bull35:244–255

Csomor PA, Yee BK, Vollenweider FX, Feldon J, Nicolet T, QuednowBB (2008) On the influence of baseline startle reactivity on theindexation of prepulse inhibition. Behav Neurosci 122:885–900

Depoortère R, Dargazanli G, Estenne-Bouhtou G, Coste A, Lanneau C,Desvignes C, Poncelet M, Heaulme M, Santucci V, Decobert M,Cudennec A, Voltz C, Boulay D, Terranova JP, Stemmelin J, RogerP, Marabout B, Sevrin M, Vigé X, Biton B, Steinberg R, FrançonD, Alonso R, Avenet P, Oury-Donat F, Perrault G, Griebel G,George P, Soubrié P, Scatton B (2005) Neurochemical, electro-physiological and pharmacological profiles of the selectiveinhibitor of the glycine transporter-1 SSR504734, a potential newtype of antipsychotic. Neuropsychopharmacology 30:1963–1985

Ellenbroek BA (2004) Pre-attentive processing and schizophrenia:animal studies. Psychopharmacology (Berl) 174:65–74

Fox GB, Esbenshade TA, Pan JB, Radek RJ, Krueger KM, Yao BB,Browman KE, Buckley MJ, Ballard ME, Komater VA, Miner H,Zhang M, Faghih R, Rueter LE, Bitner RS, Drescher KU, WetterJ, Marsh K, Lemaire M, Porsolt RD, Bennani YL, Sullivan JP,Cowart MD, Decker MW, Hancock AA (2005) Pharmacologicalproperties of ABT-239 [4-(2-{2-[(2R)-2-methylpyrrolidinyl]eth-yl}-benzofuran-5-yl) benzonitrile]: II. Neurophysiological char-acterization and broad preclinical efficacy in cognition andschizophrenia of a potent and selective histamine H3 receptorantagonist. J Pharmacol Exp Ther 313:176–190

Gaisler-Salomon I, Weiner I (2003) Systemic administration of MK-801 produces an abnormally persistent latent inhibition which isreversed by clozapine but not haloperidol. Psychopharmacology(Berl) 166:333–342

Gaisler-Salomon I, Diamant L, Rubin C, Weiner I (2008) Abnormallypersistent latent inhibition induced by MK801 is reversed byrisperidone and by positive modulators of NMDA receptorfunction: differential efficacy depending on the stage of the taskat which they are administered. Psychopharmacology (Berl)196:255–267

Gerlai R (1998) Contextual learning and cue association in fearconditioning in mice: a strain comparison and lesion study.Behav Brain Res 133:925–940

Geyer MA (2006a) The family of sensorimotor gating disorders:comorbidities or diagnostic overlaps? Neurotox Res 10:211–220

Geyer MA (2006b) Are cross-species measures of sensorimotor gatinguseful for the discovery of procognitive cotreatments forschizophrenia? Dialogues Clin Neurosci 8:9–16

Geyer MA, Braff DL (1987) Startle habituation and sensorimotorgating in schizophrenia and related animal models. SchizophrBull 13:643–668

Geyer MA, McIlwain KL, Paylor R (2002) Mouse genetic models forprepulse inhibition: an early review. Mol Psychiatry 7:1039–1053

Geyer MA, Swerdlow NR, Mansbach RS, Braff DL (1990) Startleresponse models of sensorimotor gating and habituation deficitsin schizophrenia. Brain Res Bull 25:485–498


Gould TJ, Wehner JM (1999) Genetic influences on latent inhibition.Behav Neurosci 113:1291–1296

Gray JA, Feldon J, Rawlins JNP, Hemsley DR, Smith AD (1991) Theneuropsychology of schizophrenia. Behav Brain Res 14:1–84

Gray NS, Hemsley DR, Gray JA (1992) Abolition of latent inhibitionin acute, but not chronic schizophrenics. Neurol Psychiatry BrainRes 1:83–89

Gray JA, Joseph MH, Hemsley DR, Young AM, Warburton EC,Boulenguez P, Grigoryan GA, Peters SL, Rawlins JN, Taib CT(1995) The role of mesolimbic dopaminergic and retrohippo-campal afferents to the nucleus accumbens in latent inhibition:implications for schizophrenia Behav Brain Res 71:19–31

Hince DA, Martin-Iverson MT (2005) Differences in prepulseinhibition (PPI) between Wistar and Sprague–Dawley ratsclarified by a new method of PPI standardization. BehavNeurosci 119:66–77

Hoffman HS, Searle JL (1965) Acoustic variables in the modificationof startle reaction in the rat. J Comp Physiol Psychol 60:53–58

Honey RC, Good M (1993) Selective hippocampal lesions abolish thecontextual specificity of latent inhibition and conditioning. BehavNeurosci 107:23–33

Iso H, Shimai S (1991) Running-wheel avoidance learning in mice(Mus musculus): evidence of contingency learning and differ-ences among inbred strains. J Comp Psychol 105:190–202

Javitt DC (2007) Glutamate and schizophrenia: phencyclidine, N-methyl-d-aspartate receptors, and dopamine–glutamate interac-tions. Int Rev Neurobiol 78:69–108

Killcross AS, Kiernan MJ, Dwyer D, Westbrook RF (1998) Loss oflatent inhibition of contextual conditioning following non-reinforced context exposure in rats. Q J Exp Psychol B51:75–90

Kinney GG, Sur C, Burno M, Mallorga PJ, Williams JB, Figueroa DJ,Wittmann M, Lemaire W, Conn PJ (2003) The glycinetransporter type 1 inhibitor N-[3-(4'-fluorophenyl)-3-(4'-phenyl-phenoxy) propyl]sarcosine potentiates NMDA receptor-mediatedresponses in vivo and produces an antipsychotic profile in rodentbehavior. J Neurosci 23:7586–7591

Lipina T, Labrie V, Weiner I, Roder J (2005) Modulators of theglycine site on NMDA receptors, D-serine and ALX 5407,display similar beneficial effects to clozapine in mouse models ofschizophrenia. Psychopharmacology (Berl) 179:54–67

Logue SF, Owen EH, Rasmussen DL, Wehner JM (1997) Assessmentof locomotor activity, acoustic and tactile startle, and prepulseinhibition of startle in inbred mouse strains and F1 hybrids:implications of genetic background for single gene and quanti-tative trait loci analyses. Neuroscience 80:1075–1086

Lubow RE (1989) Latent inhibition and conditioned attention theory.Cambridge University Press, New York

Lubow RE, Moore AU (1959) Latent inhibition: the effect of non-reinforced preexposure to the conditional stimulus. J CompPhysiol Psychol 66:688–694

Mackintosh NJ (1973) Stimulus selection: learning to ignore stimulithat predict no change in reinforcement. In: Hinde RA,Stevenson-Hinde J (eds) Constraints on learning. Academic,London, pp 75–96

McCaughran JJ, Mahjubi E, Decena E, Hitzemann R (1997) Genetics,haloperidol-induced catalepsy and haloperidol-induced changesin acoustic startle and prepulse inhibition. Psychopharmacology134:131–139

McNamara RK, Levant B, Taylor B, Ahlbrand R, Liu Y, Sullivan JR,Stanford K, Richtand NM (2006) C57BL/6J mice exhibit reduceddopamine D3 receptor-mediated locomotor-inhibitory functionrelative to DBA/2J mice. Neuroscience 143:141–153

Morse AC, Erwin VG, Jones BC (1993) Strain and housing affectcocaine self-selection and open-field locomotor activity in mice.Pharmacol Biochem Behav 45:905–912

Olivier B, Leahy C, Mullen T, Paylor R, Groppi VE, Sarnyai Z,Brunner D (2001) The DBA/2J strain and prepulse inhibition ofstartle: a model system to test antipsychotics. Psychopharmacol-ogy (Berl) 156:284–290

Paylor R, Baskall L, Wehner JM (1993) Behavioral dissociationsbetween C57BL/6 and DBA/2 mice on learning and memorytasks: a hippocampal dysfunction hypothesis. Psychobiology21:11–26

Paylor R, Crawley JN (1997) Inbred strain differences in prepulseinhibition of the mouse startle response. Psychopharmacology132:169–118

Paylor R, Tracy R, Wehner J, Rudy JR (1994) DBA/2 and C57BL/6mice differ in contextual fear conditioning but not auditory fearconditioning. Behav Neurosci 108:810–817

Pietropaolo S, Singer P, Feldon J, Yee BK (2008) The postweaningsocial isolation in C57BL/6 mice: preferential vulnerability in themale sex. Psychopharmacology 197:613–628

Pouzet B, Zhang WN, Weiner I, Feldon J, Yee BK (2004) Latentinhibition is spared by N-methyl-D-aspartate (NMDA)-inducedventral hippocampal lesions, but is attenuated following localactivation of the ventral hippocampus by intracerebral NMDAinfusion. Neuroscience 124:183–194

Puglisi-Allegra S, Cabib S (1997) Psychopharmacology of dopamine:the contribution of comparative studies in inbred strains of mice.Prog Neurobiol 51:637–661

Oliverio A (1967) Effects of different conditioning schedules based onvisual and acoustic conditioned stimulus on avoidance learningof two strains of mice. J Psychol 65:131–139

Ouagazzal AM, Jenck F, Moreau JL (2001) Drug-induced potentiationof prepulse inhibition of acoustic startle reflex in mice: a modelfor detecting antipsychotic activity? Psychopharmacology (Berl)156:273–283

Restivo L, Passino E, Middei S, Ammassari-Teule M (2002) Thestrain-specific involvement of nucleus accumbens in latentinhibition might depend on differences in processing configural-and cue-based information between C57BL/6 and DBA mice.Brain Res Bull 57:35–39

Sandner G, Canal NM (2007) Relationship between PPI and baselinestartle response. Cogn Neurodyn 1:27–37

Schmajuk NA, Larrauri JA (2005) Neural network model of prepulseinhibition. Behav Neurosci 119:1546–1562

Schmajuk NA, Gray JA, Lam YW (1996) Latent inhibition: a neuralnetwork approach. J Exp Psychol Anim Behav Process 22:321–349

Schiller D, Zuckerman L, Weiner I (2006) Abnormally persistentlatent inhibition induced by lesions to the nucleus accumbenscore, basolateral amygdala and orbitofrontal cortex is reversed byclozapine but not by haloperidol. J Psychiatr Res 40:167–177

Schwabe K, Freudenberg F, KochM (2007) Selective breeding of reducedsensorimotor gating in Wistar rats. Behav Genet 37:706–712

Snyder SH (1976) The dopamine hypothesis of schizophrenia: focuson the dopamine receptor. Am J Psychiatry 133:197–202

Spielewoy C, Markou A (2004) Strain-specificity in nicotineattenuation of phencyclidine-induced disruption of prepulseinhibition in mice: relevance to smoking in schizophreniapatients. Behav Genet 34:343–354

Sprott RL, Stavnes K (1975) Effects of situational variables onperformance of inbred mice in active- and passive-avoidancesituations. Psychol Rep 37:683–692

Stavnes K, Sprott RL (1975a) Effects of age and genotype onacquisition of an active avoidance response in mice. DevPsychobiol 8:437–445

Stavnes KL, Sprott RL (1975b) Genetic analysis of active avoidanceperformance in mice. Psychol Rep 36:515–521

Stevens KE, Freedman R, Collins AC, Hall M, Leonard S, Marks MJ,Rose GM (1996) Genetic correlation of inhibitory gating of


hippocampal auditory evoked response and alpha-bungarotoxin-binding nicotinic cholinergic receptors in inbred mouse strains.Neuropsychopharmacology 15:152–162

Stevens KE, Wear KD (1997) Normalizing effects of nicotine and anovel nicotinic agonist on hippocampal auditory gating in twoanimal models. Pharmacol Biochem Behav 57:869–874

Swerdlow NR, Braff DL, Geyer MA (2000) Animal models ofdeficient sensorimotor gating: what we know, what we think weknow, and what we hope to know soon. Behav Pharmacol11:185–204

Swerdlow NR, Geyer MA (1998) Using an animal model of deficientsensorimotor gating to study the pathophysiology and newtreatments of schizophrenia. Schizophr Bull 24:285–301

Swerdlow NR, Weber M, Qu Y, Light GA, Braff DL (2008) Realisticexpectations of prepulse inhibition in translational models forschizophrenia research. Psychopharmacology (Berl) 199:331–388

Tohmi M, Tsuda N, Mizuno M, Takei N, Frankland PW, Nawa H(2005) Distinct influences of neonatal epidermal growth factorchallenge on adult neurobehavioral traits in four mouse strains.Behav Genet 35:615–629

Ventura R, Alcaro A, Cabib S, Conversi D, Mandolesi L, Puglisi-Allegra S (2004) Dopamine in the medial prefrontal cortexcontrols genotype-dependent effects of amphetamine on meso-accumbens dopamine release and locomotion. Neuropsychophar-macology 29:72–80

Vetulani J, Battaglia M, Sansone M (1989) Nimodipine on shuttle-boxavoidance learning in mice: no impairment but slight improve-ment. Pharmacol Biochem Behav 56:577–581

Wagner AR (1978) Expectancies and the priming of STM. In: TigheTJ, Fowler H, Honig WK (eds) Cognitive processes in animalbehavior. Erlbaum, Hillsdale, NJ, pp 177–209

Wahlsten D (1972) Phenotypic and genetic relations between initialresponse to electric shock and rate of avoidance learning in mice.Behav Genet 2:211–240

Wehner JM, Sleight S, Upchurch M (1990) Hippocampal protein kinaseC is reduced in poor spatial learners. Brain Res 523:181–187

Weinberger SB, Koob GF, Martinez JL Jr (1992) Differences in one-way active avoidance learning in mice of three inbred strains.Behav Genet 22:177–188

Weiner I (2003) The "two-headed" latent inhibition model ofschizophrenia: modelling positive and negative symptoms andtheir treatment. Psychopharmacology (Berl) 169:257–297

Williams JH, Wellman NA, Geaney DP, Cowen PJ, Feldon J, RawlinsJNP (1998) Reduced latent inhibition in people with schizophre-nia: an effect of psychosis or of its treatment. Br J Psychiatry172:243–249

Wimer RE, Symington L, Farmer H, Schwartzkroin P (1968) Differ-ences in memory processes between inbred mouse strainsC57BL/6J and DBA/2J. J Comp Physiol Psychol 65:126–131

Wimer RE, Wimer CC, Vaugh JE, Barber RP, Balvanz BA, ChernowCR (1976) The genetic organization of neuron number inAmmon’s horns of house mouse. Brain Res 118:219–243

Yee BK, Feldon J (2009) Distinct forms of prepulse inhibitiondisruption distinguishable by the associated changes inprepulse-elicited reaction. Behav Brain Res (in press)

Yee BK, Balic E, Singer P, Schwerdel C, Grampp T, Gabernet L,Knuesel I, Benke D, Feldon J, Mohler H, Boison D (2006)Disruption of glycine transporter 1 restricted to forebrain neuronsis associated with a procognitive and antipsychotic phenotypicprofile. J Neurosci 26:3169–3181

Yee BK, Chang DL, Feldon J (2004a) The effects of dizocilpine andphencyclidine on prepulse inhibition of the acoustic startle reflexand on prepulse-elicited reactivity in C57BL6 mice. Neuro-psychopharmacology 29:1865–1877

Yee BK, Chang T, Pietropaolo P, Feldon J (2005) The expression ofprepulse inhibition of the acoustic startle reflex as a function ofthree pulse stimulus intensities, three prepulse stimulus intensi-ties, and three levels of startle responsiveness in C57BL6/J mice.Behav Brain Res 163:265–276

Yee BK, Feldon J, Rawlins JN (1995) Latent inhibition in rats isabolished by NMDA-induced neuronal loss in the retrohippo-campal region, but this lesion effect can be prevented by systemichaloperidol treatment. Behav Neurosci 109:227–240

Yee BK, Feldon J, Rawlins JN (1997) Cytotoxic lesions of theretrohippocampal region attenuate latent inhibition but spare thepartial reinforcement extinction effect. Exp Brain Res 115:247–256

Yee BK, Russig H, Feldon J (2004b) Apomorphine-induced prepulseinhibition disruption is associated with a paradoxical enhance-ment of prepulse stimulus reactivity. Neuropsychopharmacology29:240–248

Zocchi A, Orsini C, Cabib S, Puglisi-Allegra S (1998) Parallel strain-dependent effect of amphetamine on locomotor activity anddopamine release in the nucleus accumbens: an in vivo study inmice. Neuroscience 82:521–528


are dba/2 mice associated with schizophrenia-like endophenotypes? a behavioural contrast with...

Documents